Resource management method and device

Abstract

The invention provides a resource management method and device, and relates to the technical field of computer science. The method comprises the steps that portrait features of application operation state data are determined according to application index data, the application index data are obtained by conducting statistics on the application operation state data through a preset index, and the portrait features comprise at least one of an application granularity portrait, a node granularity portrait and a cluster granularity portrait; according to the portrait features, a linkage strategy is determined, and the linkage strategy represents a strategy of linkage of at least two of application vertical elasticity, application horizontal elasticity and node elasticity; and performing resource management according to the linkage strategy. According to the resource management method, multiple independent elastic strategies can be linked to cope with business requirements, and the resource management efficiency is improved.

Description

[0001] This application claims the priority of the Chinese patent application filed with the China Patent Office on July 3, 2023, with application number 202310806789.9 and application name “Resource Management Method and Device”, all contents of which are incorporated by reference in this application. Technical Field

[0002] The embodiments of the present invention relate to the field of computer science and technology, and in particular to a resource management method and device. Background Art

[0003] Computer resources are the basis for supporting the normal operation of computer-related businesses. Computer resources include, for example, computer computing resources, computer storage resources, etc. Since computer resources are limited, it is necessary to manage resources reasonably and efficiently.

[0004] For example, for cloud computing, cloud native and other scenarios, existing technologies manage resources to meet business needs by adjusting the node elasticity of the number of nodes or by adjusting the application elasticity of application resources. Node elasticity and application elasticity are independent, which results in resource management only being able to consider one of node elasticity and application elasticity. In some cases, node elasticity and application elasticity may conflict. Therefore, a more efficient resource management method that matches business needs is urgently needed. Summary of the invention

[0005] The present application provides a resource management method and device that can link multiple independent elastic strategies to meet business needs and improve resource management efficiency.

[0006] In a first aspect, the present application relates to a resource management method, comprising: determining a portrait feature of application running status data according to application indicator data, determining a linkage strategy according to the portrait feature, and performing resource management according to the linkage strategy.

[0007] The application indicator data is obtained by statistically analyzing the application running status data using preset indicators, and the portrait features include at least one of an application granularity portrait, a node granularity portrait, and a cluster granularity portrait.

[0008] The running of an application consumes computer resources, and the application also generates status data based on the consumption and dependency of computer resources. This status data is the application running status data.

[0009] The preset indicators can be understood as parameter indicators used to evaluate the operation of the application. The application indicator data obtained by using the preset indicators to collect statistics on the application operation status data can be used as the parameter value for evaluating the operation of the application. Therefore, the portrait features of the application operation status data determined based on the application indicator data can characterize the characteristics of the application operation.

[0010] The application granularity portrait, node granularity portrait and cluster granularity portrait represent the portrait features of application granularity, node granularity and cluster granularity respectively.

[0011] The linkage strategy represents a strategy for linking at least two of application vertical elasticity, application horizontal elasticity, and node elasticity.

[0012] According to the resource management method of the embodiment of the present application, multiple application vertical elasticity, application horizontal elasticity and node elasticity can be linked to perform resource management. In the case of linking the above-mentioned multiple independent elasticity strategies, the resource management method of the embodiment of the present application, because the application operation consumes computer resources, uses the image characteristics of the application operation status data as the basis for determining the linkage strategy, which can more accurately match the function of the computer resources. The linkage strategy determined according to the image characteristics can unify the linkage of multiple application vertical elasticity, application horizontal elasticity and node elasticity to cope with fluctuations in business demand, especially the "business tide" phenomenon in which the business has a regular fluctuation pattern, and also has higher resource management efficiency. Higher efficiency of resource management is reflected, for example, in the case of linking multiple independent elasticity strategies, the resource management methods are richer, the resource management is more reasonable, and there will be no conflicts in elasticity strategies.

[0013] In a possible embodiment, the application running status data includes at least one of running indicator data and application characteristic data.

[0014] The operation indicator data includes at least one of the following: storage resource occupancy data, computing resource occupancy data, event transmission rate, and traffic processing rate.

[0015] The storage resource occupancy data represents the storage resources required for the application to run, and the storage resources include, for example, memory. The computing power resource occupancy data represents the computing power resources required for the application to run, and the computing power resources include, for example, CPU computing power resources and GPU computing power resources.

[0016] The application characteristic data includes at least one of the following: an operation sensitivity coefficient, a storage sensitivity coefficient, and a delay sensitivity coefficient.

[0017] The computing sensitivity coefficient can be understood as the change in computing power caused by a one-unit change in computing power resources. The storage sensitivity coefficient can be understood as the change in storage capacity caused by a one-unit change in storage resources. The delay sensitivity coefficient can be understood as the change in delay duration caused by a one-unit change in computing power resources.

[0018] The computer resources required for application operation are used for data storage and calculation, etc. According to the resource management method of the embodiment of the present application, in addition to meeting the demand for computer resources to ensure normal processing of business, computer resource management can also evaluate the resource management scheme on this basis to select a better and more efficient resource management scheme. For example, operation indicator data such as storage resource occupancy data and computing power resource occupancy data can be used to evaluate the demand for computer resources for normal processing of business. For example, application feature data can be used to evaluate the efficiency of computer resource use. For example, a higher calculation sensitivity coefficient indicates that the gain of increasing unit computing power resources is higher, so the utilization rate of resources is higher, and resource management is more reasonable and efficient.

[0019] In a possible example, performing resource management according to the linkage strategy includes: adjusting at least one of the following according to the linkage strategy to perform resource management: the number of nodes, the specification of the resource deployment unit, and the number of resource deployment units.

[0020] The specification of the resource deployment unit, the number of resource deployment units, and each node are related to resources.

[0021] The resource deployment unit is deployed on a node, and the cluster includes at least one node.

[0022] Adjusting the number of nodes is called node elasticity, adjusting the specifications of resource deployment units is called application vertical elasticity, and adjusting the number of resource deployment units is called application horizontal elasticity. In this way, multiple of application vertical elasticity, application horizontal elasticity, and node elasticity can be linked to perform resource management.

[0023] In a possible example, determining a linkage strategy according to the portrait features may include: determining a linkage strategy that satisfies a target rule according to the portrait features.

[0024] The target rule includes at least one of a service guarantee rule and a resource consumption minimum rule.

[0025] The linkage strategy that satisfies the rule of minimum resource consumption indicates that the linkage strategy has the minimum resource consumption among the optional linkage strategies, thereby improving resource utilization efficiency.

[0026] In the case where the target rule includes a service assurance rule and a resource consumption minimization rule, the priority of the service assurance rule is higher than the priority of the resource consumption minimization rule.

[0027] According to the resource management method of the embodiment of the present application, by determining the linkage strategy that meets the target rules based on the portrait characteristics, it can at least ensure that the normal operation of the business application is less affected or the resource consumption is less, so as to protect the user experience or improve the resource utilization efficiency.

[0028] In a possible example, the portrait feature includes an application granularity portrait. Determining the linkage strategy based on the portrait feature may include: determining a strategy for linking application vertical elasticity with application horizontal elasticity based on the application granularity portrait.

[0029] The application granularity portrait is a portrait feature of the application running status data measured in the application dimension. Application vertical elasticity and application horizontal elasticity are also two different elasticity strategies in the application dimension. Compared with node elasticity, the costs of application vertical elasticity and application horizontal elasticity are lower. Therefore, according to the resource management method of the embodiment of the present application, based on the application granularity portrait in the application dimension, a strategy for linking application vertical elasticity and application horizontal elasticity in the application dimension can be determined to achieve unified linkage of elasticity in the application dimension, which can reduce cost and improve resource management efficiency.

[0030] In a possible example, the portrait feature includes a node granularity portrait. Determining the linkage strategy based on the portrait feature may include: determining a strategy for linking application vertical elasticity with application horizontal elasticity based on the application granularity portrait and the node granularity portrait.

[0031] In some cases, application vertical elasticity and application horizontal elasticity require that node resources meet conditions. For example, if application horizontal elasticity requires increasing the number of Pods or application vertical elasticity requires increasing the resource specifications of Pods, the current node's resources must be able to accommodate the increased number of Pods or the current node's resources must meet the specifications of the added Pods.

[0032] According to the resource management method of the embodiment of the present application, the node granularity portrait is also used as the basis for determining the linkage strategy between application vertical elasticity and application horizontal elasticity, which can cope with the above situation and more accurately determine the reasonable linkage strategy between application vertical elasticity and application horizontal elasticity.

[0033] In one possible example, the portrait feature includes a cluster granularity portrait. Determining the linkage strategy based on the portrait feature may include: determining the node elasticity and application vertical elasticity and / or application horizontal elasticity linkage strategy based on the application granularity portrait and the cluster granularity portrait and / or the node granularity portrait.

[0034] The node granularity portrait is a portrait feature of the application running status data measured from the node dimension, and the cluster granularity portrait is a portrait feature of the application running status data measured from the cluster dimension. The cluster includes at least one node. Therefore, the linkage strategy including the node elasticity strategy can be determined at least based on the node granularity portrait and / or the cluster granularity portrait, thereby unifying and linking multiple independent elasticity strategies to manage resources more efficiently.

[0035] In a possible example, determining the linkage strategy according to the portrait features may include: processing the portrait features using a model to obtain the linkage strategy.

[0036] A model can be understood as a function that receives data as input and generates output. A model may include a machine learning model, for example.

[0037] The model can be pre-trained, for example, the input of the model is the image feature, and the output of the model is the linkage strategy. Model training adjusts the model parameters. By adjusting the model parameters, the performance of the model will also change. When the model meets the requirements, the model training is completed. Therefore, the trained model can meet the predetermined requirements in terms of performance such as accuracy. Therefore, according to the resource management method of the embodiment of the present application, the linkage strategy output by the model at least meets the performance requirements such as accuracy requirements, which is a more intelligent and efficient way to determine the linkage strategy.

[0038] In a possible example, using a model to process the image features to obtain a linkage strategy may include: using at least one first model to process the image features to obtain at least one initial linkage strategy corresponding to each first model. According to the at least one initial linkage strategy, a target linkage strategy is determined.

[0039] When there are plural first models, model parameters of any plural first models are different from each other.

[0040] The first model is obtained by training with training samples. The model parameters of the first model are different before and after training, and the performance of the first models with different model parameters is also different. The portrait features are processed by multiple first models to obtain multiple initial linkage strategies and determine the target linkage strategy based on the multiple initial linkage strategies. The advantages of each first model can be combined to avoid inaccurate linkage strategies caused by the defects of a certain first model. Therefore, the resource management method according to the embodiment of the present application can determine a more accurate and reasonable linkage strategy, thereby correspondingly improving the resource management efficiency.

[0041] In a possible example, when there are multiple first models, the multiple first models are trained according to training samples of different business scenarios, so that model parameters of each first model are different.

[0042] The first model is trained based on the training samples, and the first models trained by the training samples of different business scenarios are different. For example, the first model trained by the training samples of a certain business scenario better learns the characteristics of this business scenario, so the performance of the first model is better for the input data of the business scenario. If the first model is not trained by the training samples of another business scenario, the performance of the first model for the input data of the other business scenario is poor.

[0043] Due to the diversity and complexity of business scenarios, it is difficult to cope with all business scenarios through a certain first model and to achieve excellent performance in all business scenarios. Therefore, according to the resource management method of the embodiment of the present application, multiple first models obtained by training with training samples of different business scenarios can be used to distinguish different application scenarios, so as to accurately determine the linkage strategy and thereby improve resource management efficiency.

[0044] In a possible example, determining a target linkage strategy based on at least one initial linkage strategy may include: using a second model to process at least one initial linkage strategy and portrait features and / or application indicator data to obtain a target linkage strategy.

[0045] In the case where there are multiple initial linkage strategies, it is necessary to determine a target linkage strategy from the multiple initial linkage strategies. This process can be understood as analyzing multiple candidate initial linkage strategies to determine the target linkage strategy.

[0046] When it is necessary to comprehensively analyze and compare each initial linkage strategy to determine the target linkage strategy, it is relatively simple to use only the initial linkage strategy as a basis. Therefore, on this basis, the target linkage strategy can be comprehensively and accurately determined based on at least one initial linkage strategy and portrait features and / or application indicator data.

[0047] In a possible example, determining the linkage strategy according to the portrait features may include: determining application indicator data at a future time according to the portrait features. Determining the requirement specification and the required quantity of the resource deployment unit according to the application indicator data at the future time. Determining the linkage strategy according to the requirement specification and the required quantity of the resource deployment unit.

[0048] When the application is running, the application indicator data is dependent in the time dimension. For example, the long short-term memory network can be used to determine the application indicator data at the future moment based on the portrait features. The long short-term memory network is a time recursive neural network with long-term memory capabilities. The network structure of the long short-term memory network includes one or more neurons with forgettable and memory functions, which can be used to selectively "memorize" the output results of the previous neuron. Therefore, the long short-term memory network can adapt to the characteristics of the application indicator data being dependent in the time dimension, thereby accurately predicting the application indicator data at the future moment.

[0049] A resource deployment unit can be understood as a unitized resource deployment instance. Each resource deployment unit can be deployed independently, and the resources and environment of each resource deployment unit are shared.

[0050] The portrait features of the application running status data represent the characteristics of the application running. Based on the portrait features, the application index data at the future moment can be predicted. Based on the application index data at the future moment, the demand specifications and demand quantities of the resource deployment unit can be accurately determined, and then a reasonable and efficient linkage strategy can be determined. The demand specifications and demand quantities of the resource deployment unit can be understood as the resource demand at the granularity of the resource deployment unit that matches the future application index data. The resource demand is represented by the demand specifications and demand quantities of the resource deployment unit.

[0051] In a possible example, determining the required specifications and required quantity of the resource deployment unit according to the application indicator data at a future moment may include: determining a first specification and a first quantity of the resource deployment unit that matches the application indicator data at a historical moment. Determining a second specification and a second quantity of the resource deployment unit that matches the application indicator data at a future moment. Determining the required specifications and required quantity of the resource deployment unit according to the first specification and the first quantity and the second specification and the second quantity.

[0052] The application indicator data at the historical moment is obtained by statistically analyzing the application running status data at the historical moment using preset indicators.

[0053] Exemplarily, for example, based on the application indicator data at the historical moment, the first specification and the first quantity of the resource deployment units matching the application indicator data at the historical moment may be determined.

[0054] The following situation exists: the first specification and the first quantity corresponding to the application indicator data at the historical moment indicate the need for expansion, while the second specification and the second quantity corresponding to the application indicator data at the future moment indicate the need for reduction. In this case, the resources released by the future reduction can offset at least part of the resources required for the historical expansion. The offset part can be determined based on the first specification and the first quantity and the second specification and the second quantity. The resources that actually need to be allocated are, for example, the required specifications and the required quantities of the resource deployment unit.

[0055] According to the resource management method of the embodiment of the present application, based on the dependency characteristics of resource management in the time dimension, the complementary resource demands of historical moments and future moments are offset, so that a more reasonable resource management plan can be determined and resource management efficiency can be improved.

[0056] For example, in order to more closely connect the resource requirements of historical moments and future moments in the time dimension to improve resource management efficiency, the span between historical moments and future moments can be set to be smaller. For example, the historical moment can be the current moment, and the future moment can be the next moment of the current moment.

[0057] In a possible example, determining the portrait feature of the application running status data according to the application indicator data may include: determining at least one of the relationship between the application indicator data and resources, the topological relationship between resource deployment units, and the available resources of the node according to the application indicator data. Determine the portrait feature according to at least one of the relationship between the application indicator data and resources, the topological relationship between resource deployment units, and the available resources of the node.

[0058] Exemplarily, for example, the relationship between application indicator data and resources, the topological relationship between resource deployment units, and at least one of the available resources of a node may be used as a portrait feature.

[0059] The profile features of the application running status data represent the features of the application running, and the profile features of the application running status data are subsequently used to allocate resources. The relationship between the application indicator data and resources can represent the correlation between resource usage caused by the application running, so it can be used as a dimension to evaluate the features of the application running (i.e., the profile features), and can be used to assist in resource management later.

[0060] Since the resource management method of the embodiment of the present application can perform resource management at the granularity of resource deployment units, for example, resource management can be performed by adjusting the resource specifications and quantity of resource deployment units, and the topological relationship between resource deployment units can characterize the overall relationship of resource deployment units, the topological relationship between resource deployment units can be used as the basis for resource management, making the resource management of the embodiment of the present application more reasonable and efficient.

[0061] The available resources of a node can be understood as the resources of a node that can be used for resource management. The resource management method of the embodiment of the present application, when performing resource management through the application dimension, still needs to be based on the premise that the node resources meet the conditions. In some cases, a part of the node resources is used to maintain the basic functions of the node, etc. If this part of the resources is used for allocation, it will affect the basic functions. At this time, the available resources of the node can be obtained by the difference between the total resources of the node and the resources required to maintain the basic functions. The available resources of the node can accurately represent the resources that can actually be used for resource management. For example, accurate and efficient resource management can be performed based on the available resources of the node in the future.

[0062] In a possible example, determining the requirement specifications and required quantity of resource deployment units according to the application indicator data at the future moment may include: determining a target resource quantity matching the application indicator data at the future moment. Determining the requirement specifications and required quantity of resource deployment units according to the target resource quantity.

[0063] The target resource volume is determined based on the relationship between application indicator data and resources.

[0064] The resource management method of the embodiment of the present application utilizes the service provided by the resource deployment unit to process application indicator data: based on the relationship between the application indicator data and the resources, the target resource quantity that matches the application indicator data at the future moment can be determined, and the target resource quantity can be allocated to the granularity of the resource deployment unit, that is, the required specifications and required quantity of the resource deployment unit are determined according to the target resource quantity. For example, the specifications and quantity of the resource deployment unit can be adjusted according to the required specifications and required quantity of the resource deployment unit, so that the resource deployment unit after resource management matches the application indicator data at the future moment. Therefore, the resource deployment unit after resource management has the ability to cope with the application indicator data at the future moment.

[0065] In a possible example, determining the target resource quantity matching the application indicator data at the future moment may include: determining at least one resource quantity data matching the application indicator data at the future moment according to the relationship between the application indicator data and the resource, and determining the target resource quantity with the best cost performance from the at least one resource quantity data.

[0066] The best price / performance ratio is determined based on the minimum value of the ratio between resources and application indicator data.

[0067] In the case where the relationship between application indicator data and resources is one-to-many, multiple resource quantity data can be determined based on the application indicator data of a certain value at a future moment. According to the resource management method of the embodiment of the present application, the target resource quantity can be determined from multiple resource quantity data by the parameter of the best cost performance ratio. The ratio between resources and application indicator data can characterize the resources required for each processing of an application indicator data. The minimum value of the ratio between resources and application indicator data indicates that the resources required for each processing of an application indicator data are the minimum, that is, the resource utilization rate is higher. Therefore, the resource utilization rate corresponding to the target resource quantity determined by the resource management method of the embodiment of the present application is higher, and the resource management efficiency is also higher.

[0068] In a possible example, adjusting the specification of the resource deployment unit and / or the number of resource deployment units for resource management may include: calling resources of a target node in the candidate nodes to adjust the specification of the resource deployment unit and / or the number of resource deployment units.

[0069] The resources of the target node meet the resource requirements of the specification of the resource deployment unit and / or the number of resource deployment units, and the remaining resources of the target node after the resources are called are the smallest among the candidate nodes.

[0070] In the case of calling node resources for application-dimensional elasticity, there may be multiple candidate nodes available for resource scheduling. According to the resource management method of the embodiment of the present application, since the resourceization of the target node meets the resource requirements of the specifications of the resource deployment unit and / or the number of resource deployment units, and the remaining resources of the target node after being called are the smallest among the candidate nodes, this indicates that the target node can also achieve the maximum node resource utilization efficiency under the premise of meeting the resource requirements of the specifications of the resource deployment unit and / or the number of resource deployment units. Therefore, resource management is more efficient by calling the resources of the target node in the candidate node to adjust the specifications of the resource deployment unit and / or the number of resource deployment units, and the resource utilization rate is also higher.

[0071] In a possible example, performing resource management according to the linkage policy may include: determining a linkage policy instruction according to the linkage policy.

[0072] The linkage policy instruction is used to instruct the execution of the linkage policy to complete resource management.

[0073] Exemplarily, for example, the executor may respond to the received linkage strategy instruction and execute the linkage strategy to complete resource management.

[0074] According to the resource management method of the embodiment of the present application, the determination of the linkage strategy and the execution of the linkage strategy can be decoupled. After the linkage strategy is determined, the linkage strategy instruction instructs the execution of the linkage strategy. For example, it supports separately updating the determination process of the linkage strategy and the execution process of the linkage strategy, which is more flexible.

[0075] In a second aspect, the present application relates to a resource management device, comprising: a portrait feature determination module, a linkage strategy determination module and a resource management module.

[0076] The portrait feature determination module is used to determine the portrait features of the application running status data based on the application indicator data, wherein the application indicator data is obtained by statistically analyzing the application running status data using preset indicators, and the portrait features include at least one of the application granularity portrait, the node granularity portrait and the cluster granularity portrait.

[0077] The linkage strategy determination module is used to determine the linkage strategy according to the portrait features, wherein the linkage strategy represents a strategy for linking at least two of application vertical elasticity, application horizontal elasticity, and node elasticity.

[0078] The resource management module is used to manage resources according to linkage strategies.

[0079] In a possible example, the linkage strategy determination module includes: a linkage strategy first determination sub-module, which is used to determine a linkage strategy that meets a target rule based on a portrait feature, wherein the target rule includes at least one of a business assurance rule and a minimum resource consumption rule, and wherein, when the target rule includes a business assurance rule and a minimum resource consumption rule, the priority of the business assurance rule is higher than the priority of the minimum resource consumption rule.

[0080] In a possible example, where the portrait feature includes an application granularity portrait, the linkage strategy determination module includes: a linkage strategy second determination submodule, which is used to determine a strategy for linking application vertical elasticity and application horizontal elasticity based on the application granularity portrait.

[0081] In a possible example, where the portrait feature includes a node granularity portrait, the linkage strategy determination module includes: a linkage strategy third determination submodule, used to determine a strategy for linking application vertical elasticity with application horizontal elasticity based on the application granularity portrait and the node granularity portrait.

[0082] In one possible example, in which the portrait feature includes a cluster granularity portrait, the linkage strategy determination module includes: a fourth linkage strategy determination submodule, used to determine the node elasticity and application vertical elasticity and / or application horizontal elasticity linkage strategy based on the application granularity portrait and the cluster granularity portrait and / or the node granularity portrait.

[0083] In a possible example, the linkage strategy determination module includes: a linkage strategy fifth determination submodule, which is used to process the portrait features using a model to obtain a linkage strategy.

[0084] In a possible example, the fifth linkage strategy determination submodule includes: an initial linkage strategy determination unit and a target linkage strategy determination unit.

[0085] The initial linkage strategy determination unit is used to process the portrait features using at least one first model to obtain at least one initial linkage strategy corresponding to each first model, wherein when there are multiple first models, the model parameters of any multiple first models are different from each other.

[0086] The target linkage strategy determining unit is used to determine a target linkage strategy according to at least one initial linkage strategy.

[0087] In a possible example, when there are multiple first models, the multiple first models are trained based on training samples of different business scenarios, so that model parameters of each first model are different.

[0088] In a possible example, the target linkage strategy determination unit includes: a target linkage strategy determination subunit, which is used to use the second model to process at least one initial linkage strategy and portrait features and / or application indicator data to obtain a target linkage strategy.

[0089] In a possible example, the linkage strategy determination module includes: a prediction submodule, a demand resource determination submodule, and a linkage strategy sixth determination submodule.

[0090] The prediction submodule is used to determine the application indicator data at future moments based on the portrait features.

[0091] The demand resource determination submodule is used to determine the demand specifications and quantity of resource deployment units based on the application indicator data at a future moment.

[0092] The sixth linkage strategy determination submodule is used to determine the linkage strategy according to the demand specifications and demand quantities of the resource deployment units.

[0093] In a possible example, the required resource determination submodule includes: a first matching unit, a second matching unit, and a required resource determination unit.

[0094] The first matching unit is used to determine a first specification and a first quantity of resource deployment units that match application indicator data at a historical moment, wherein the application indicator data at a historical moment is obtained by statistically analyzing application running status data at a historical moment using preset indicators.

[0095] The second matching unit is used to determine a second specification and a second quantity of resource deployment units that match the application indicator data at a future moment.

[0096] The demand resource determination unit is used to determine the demand specification and the demand quantity of the resource deployment unit according to the first specification and the first quantity and the second specification and the second quantity.

[0097] In a possible example, the portrait feature determination module includes: a portrait feature dimension determination submodule and a portrait feature determination submodule.

[0098] The portrait feature dimension determination submodule is used to determine at least one of the relationship between application indicator data and resources, the topological relationship between resource deployment units, and the available resources of the node based on the application indicator data.

[0099] The portrait feature determination submodule is used to determine the portrait feature based on at least one of the relationship between application indicator data and resources, the topological relationship between resource deployment units, and the available resources of the node.

[0100] In a possible example, the required resource determination submodule includes: a target resource quantity determination unit and a required resource determination unit.

[0101] The target resource amount determination unit is used to determine the target resource amount that matches the application indicator data at a future moment, wherein the target resource amount is determined based on the relationship between the application indicator data and the resources.

[0102] The demand resource determination unit is used to determine the demand specifications and demand quantities of the resource deployment unit according to the target resource quantity.

[0103] In a possible example, the target resource amount determination unit includes: a resource matching subunit and a target resource amount determination subunit.

[0104] The resource matching subunit is used to determine at least one resource quantity data matching the application indicator data at a future moment according to the relationship between the application indicator data and the resources.

[0105] The target resource quantity determination subunit is used to determine the target resource quantity with the best cost performance ratio from at least one resource quantity data, wherein the best cost performance ratio is determined based on the minimum value of the ratio between the resource and the application indicator data.

[0106] In one possible example, the resource management module includes: an adjustment submodule, which is used to adjust at least one of the following according to the linkage strategy to perform resource management: the number of nodes, the specifications of the resource deployment units, and the number of resource deployment units, wherein the specifications of the resource deployment units, the number of resource deployment units, and each node are related to resources, the resource deployment units are deployed on the nodes, and the cluster includes at least one node.

[0107] In one possible example, the adjustment submodule includes: a node resource calling unit, used to call the resources of the target node among the candidate nodes to adjust the specifications of the resource deployment units and / or the number of resource deployment units, wherein the resources of the target node meet the resource requirements of the specifications of the resource deployment units and / or the number of resource deployment units, and the remaining resources of the target node after the resources are called are the smallest among the candidate nodes.

[0108] In a possible example, the resource management module further includes: a linkage policy instruction determination submodule, which is used to determine a linkage policy instruction according to the linkage policy, wherein the linkage policy instruction is used to instruct execution of the linkage policy to complete resource management.

[0109] In one possible example, the application indicator data includes at least one of operation indicator data and application characteristic data, wherein the operation indicator data includes at least one of the following: storage resource occupancy data, computing resource occupancy data, event transmission rate, and traffic processing rate, and the application characteristic data includes at least one of the following: operation sensitivity coefficient, storage sensitivity coefficient, and delay sensitivity coefficient.

[0110] In a third aspect, the present application relates to an electronic device, comprising: at least one processor; the at least one processor is used to execute instructions so that the electronic device executes the above-mentioned resource management method.

[0111] In some embodiments, the instructions are stored in a memory that is communicatively connected or coupled to at least one processor.

[0112] In a fourth aspect, a computer-readable storage medium storing computer instructions is provided, wherein when the computer instructions are executed, the computer executes the resource management method described above. In some embodiments, the computer-readable storage medium is a non-transitory storage medium.

[0113] In a fifth aspect, the present application relates to a computer program product, including a computer program, wherein the computer program is stored on a readable storage medium, and when the computer program is executed, the computer implements the above-mentioned resource management method. BRIEF DESCRIPTION OF THE DRAWINGS

[0114] The following is an introduction to the drawings used in the embodiments of the present application.

[0115] Figure 1A A schematic diagram showing application level elasticity is shown schematically;

[0116] Figure 1B A schematic diagram showing the application of vertical elasticity is shown;

[0117] Figure 1C A schematic diagram schematically illustrates node resilience;

[0118] Figure 2A A system architecture of a resource management method and device according to an embodiment of the present disclosure is schematically shown;

[0119] Figure 2B Another system architecture of the resource management method and device according to an embodiment of the present disclosure is schematically shown;

[0120] Figure 3 A flowchart of a resource management method according to an embodiment of the present disclosure is schematically shown;

[0121] Figure 4 A schematic diagram schematically shows a resource allocation method to a resource deployment unit according to an embodiment of the present disclosure;

[0122] Figure 5 A schematic diagram schematically shows a resource management method according to another embodiment of the present disclosure, which determines a strategy for linking application vertical elasticity and application horizontal elasticity based on an application granularity portrait and a node granularity portrait;

[0123] Figure 6 A schematic diagram schematically shows a strategy for determining node elasticity and application vertical elasticity and / or application horizontal elasticity linkage according to an application granularity portrait and a cluster granularity portrait and / or a node granularity portrait according to a resource management method according to another embodiment of the present disclosure;

[0124] Figure 7 The schematic diagram schematically shows a resource management method according to another embodiment of the present disclosure when the application running status data is stored in a database;

[0125] Figure 8 A schematic diagram schematically shows a resource management method according to another embodiment of the present disclosure;

[0126] Fig. 9 A block diagram of a resource management device according to an embodiment of the present disclosure is schematically shown;

[0127] Fig.10 The block diagram schematically shows an electronic device that can implement the resource management method of the embodiment of the present disclosure. DETAILED DESCRIPTION

[0128] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0129] The term "and / or" in this article is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone.

[0130] The terms "first" and "second" in the description and claims of the embodiments of the present application are used to distinguish different objects rather than to describe a specific order of objects. For example, a first target object and a second target object are used to distinguish different target objects rather than to describe a specific order of target objects.

[0131] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0132] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0133] The detailed background of the resource management method according to the embodiment of the present disclosure will be described in detail below.

[0134] Computer resources are the basis for supporting the normal operation of computer-related businesses. Computer resources include, for example, computer computing resources, computer storage resources, etc. Computer computing resources may include, for example, CPU computing resources (CPU stands for Central Processing Unit, central processing unit), GPU computing resources (GPU stands for Graphic Processing Unit, graphics processing unit), etc. Since computer resources are limited, it is necessary to manage resources reasonably and efficiently.

[0135] Cloud computing is an Internet-based computing model that provides computer resources and services (such as software, development platforms, and networks) to users for use and delivery in an on-demand, elastic, and scalable manner. Cloud computing transfers computing power and storage capacity from local computers to cloud data centers. Users do not need to own and manage actual physical devices, and can use services and store data through the Internet.

[0136] Cloud Native is a general term for a type of technology for building applications. Cloud Native technology can be used to build elastically scalable applications that can be run in different environments.

[0137] Cloud native involves technologies such as microservices, containers, serverless architecture, service network, cloud, DevOps, etc. Microservices can decouple applications and realize independent development and deployment of each function. Containers are a lightweight virtualization technology, which is essentially a series of processes separated from the rest of the system. The "serverless" meaning of serverless architecture is no longer focusing on the underlying service architecture, but focusing on business logic. Service network can be regarded as more intelligent proxy software. Based on containers and microservices, service network allows users to manage communication between services more finely and intelligently. Cloud is a network that provides resources and is the foundation of cloud native technology. DevOps (a combination of Development and Operations) is a collective name for a set of processes, methods and systems used to promote communication, collaboration and integration between development (application / software engineering), technical operations and quality assurance (QA) departments.

[0138] With the rapid development of cloud computing and cloud-native computing, and the widespread application of cloud-native technologies in business, there is no longer an upper limit on the capacity of application processing services. However, it also faces a relatively large problem: as the business application load increases, how to trigger the horizontal elasticity and vertical elasticity of the application on demand according to the characteristics of the business application, as well as the elasticity of the trigger node.

[0139] Taking cloud computing and cloud native scenarios as examples, some implementation methods perform resource management based on application elasticity or node elasticity. The following will take the management of Pod as a resource deployment unit for resource management as an example. Pod is the smallest scheduling unit in Kubernetes (Kubernetes is an open source containerized application for managing multiple hosts in a cloud platform. The goal of Kubernetes is to make the deployment of containerized applications simple and efficient. Kubernetes provides a mechanism for application deployment, planning, updating, and maintenance. Kubernetes is also referred to as K8S). A Pod encapsulates a container (or multiple containers), and the containers in the Pod share storage, network, etc. The entire Pod can be regarded as a virtual machine, and each container of the Pod can be understood as a process running on the virtual machine. All containers in the same Pod are uniformly arranged and scheduled.

[0140] Application elasticity can include: horizontal pod autoscaler (HPA) and vertical pod autoscaler (VPA).

[0141] Take the resource deployment unit of managing Pod as an example. Figure 1A A schematic diagram of applying horizontal elasticity is shown schematically.

[0142] like Figure 1A As shown, applying the horizontal elastic module (i.e. Figure 1A The HPA shown in the figure) periodically obtains the information from the indicator monitoring module (i.e. Figure 1A The Metrics Server used by the Pod Management Module (i.e., the Pod Management Module) performs metric queries, obtains scaling metrics, and calculates the expected number of replicas (i.e., the number of Pods). Figure 1A The Deployment shown in the figure is expanded or reduced according to the expected number of replicas, so that the number of Pods can bear the business pressure.

[0143] exist Figure 1A In the diagram shown, the number of replicas before scaling based on application-level elasticity is 2, corresponding to the two Pods Pod-1 and Pod-2. Figure 1A The schematic diagram shows an example of capacity expansion. The number of replicas after that is N (N is a positive integer greater than 2), corresponding to a total of N Pods from Pod-1 to Pod-N.

[0144] Take the resource deployment unit of managing Pod as an example. Figure 1B A schematic diagram showing the application of vertical elasticity.

[0145] like Figure 1B As shown, the resource recommendation module (i.e. Figure 1B The Recommender shown in the figure is periodically updated from the indicator monitoring module (i.e. Figure 1B Metrics Server) to query the metrics, obtain the scaling metrics, and apply the vertical elastic module (i.e. Figure 1B The VPA shown in the figure is configured with application elasticity. The resource recommendation module reads the application elasticity configuration from the application vertical elasticity module and calculates the resource recommendation, and then provides the calculated resource recommendation to the application vertical elasticity module. The application vertical elasticity module sends the Pod resource recommendation to the update module (i.e. Figure 1B The update module is updated according to the resource configuration required by the Pod (i.e., the Pod resource recommendation). Figure 1B In the example, for example, if the Pod resource recommendation indicates that the Pod needs to be terminated, the Pod management module (i.e. Figure 1B The Deployment shown in the figure) creates a Pod, and the admission controller (i.e. Figure 1B The Admission Controller shown in the figure obtains resource recommendations and increases the CPU computing resources by 250 microcores, so that the computing resources of the Pod are adjusted from 250 microcores to 500 microcores. After the computing resources of the Pod are adjusted, the Pod can handle more business.

[0146] Take the resource deployment unit of managing Pod as an example. Figure 1C A schematic diagram showing node resilience is shown schematically.

[0147] like Figure 1C As shown, the node elasticity module (Cluster AutoScaler, Figure 1C The CA shown in the figure monitors the new Pods to be scheduled in the cluster, simulates the scheduling, calculates the number of nodes to be expanded, and then selects nodes of appropriate specifications and calls the interface of the cloud provider (the cloud provided by the cloud provider) to expand the capacity. Figure 1C In the example, a new node N is initiated, so that the number of nodes changes from 2 (the existing nodes are node 1 and node 2) to 3. Scaling down is a reverse process, which starts with scaling down the Pods, expelling the Pods, and finally scaling down and deleting the idle nodes.

[0148] Some implementations utilize one of application horizontal elasticity, application level elasticity, and node elasticity for resource management. It is not possible to link any of application horizontal elasticity, application level elasticity, and node elasticity, resulting in low efficiency in resource management and resource utilization to match business needs.

[0149] The embodiment of the present disclosure provides a resource management method to improve the traditional resource management method.

[0150] Figure 2A and Figure 2B The schematic diagrams schematically show the system architecture of the resource management method and device according to the embodiments of the present disclosure. It should be noted that: Figure 2A and Figure 2B It is only an example of a system architecture to which the resource management method and apparatus of the embodiments of the present disclosure can be applied, so as to help those skilled in the art understand the technical content of the present disclosure, but it does not mean that the embodiments of the present disclosure cannot be applied to other devices, systems, environments or scenarios.

[0151] Figure 2A A system architecture of a resource management method and device according to an embodiment of the present disclosure is schematically shown.

[0152] like Figure 2A As shown, the system architecture 200 -A in the embodiment of the present disclosure may include: terminal devices 201 , 202 , 203 , a network 204 and a server 205 .

[0153] The terminal devices 201 , 202 , and 203 may be various electronic devices with a display and a processor, including but not limited to a smart phone, a tablet computer, a laptop computer, a desktop computer, and the like.

[0154] The server 205 may be a server that provides various services. The server 205 may have a cloud computing function, that is, the server may include a cloud server or be a part of a cloud server.

[0155] The network 204 is used to provide a medium for communication links between the terminal devices 201, 202, and 203. The network 204 may include various connection types, such as wired or wireless communication links or optical fiber cables.

[0156] Exemplarily, the application indicator data or the application running status data may be generated on the terminal device, for example, the server may receive the application indicator data from the terminal device and determine the image characteristics of the application status data according to the application indicator data. According to the image characteristics, the linkage strategy is determined. According to the linkage strategy, resource management is performed.

[0157] Figure 2B Another system architecture of the resource management method according to an embodiment of the present disclosure is schematically shown.

[0158] like Figure 2B As shown, Figure 2A Different from the system architecture shown, the system architecture 200-B of the embodiment of the present disclosure is based on cloud computing or cloud native architecture, and may include, for example, a cluster consisting of multiple nodes. Figure 2BIn the example of FIG. 2 , a specific example is schematically shown in which the cluster includes a total of 3 nodes, namely, node N1 to node N3 , and the nodes communicate and interact with each other through a network 204 .

[0159] Each node may be, for example, a computer device, a terminal device, a server, etc. Application indicator data or application status data may be generated at the node, and the node or cluster may determine the image characteristics of the application status data based on the application indicator data. Based on the image characteristics, a linkage strategy may be determined. Based on the linkage strategy, resource management may be performed.

[0160] It should be noted that the resource management method provided in the embodiment of the present disclosure can be executed by a node, a cluster or a server. Accordingly, the resource management device provided in the embodiment of the present disclosure can be set in a node, a cluster or a server.

[0161] The number of terminal devices 201, 202, 203, networks 204, servers 205, nodes and clusters in the above embodiments are only for illustration, and any number of terminal devices 201, 202, 203, networks 204, servers 205, nodes and clusters can be set as required.

[0162] The present disclosure provides a resource management method. Figure 2A and Figure 2B The system architecture of Figures 3 to 8 The resource management method of the exemplary embodiment of the present disclosure is described below. The resource management method of the exemplary embodiment of the present disclosure can be, for example, Figure 2A Alternatively, it can be executed by the server shown in Figure 2B Execute on the indicated node or cluster.

[0163] Figure 3 The flowchart of a resource management method 300 according to an embodiment of the present disclosure is schematically shown.

[0164] like Figure 3 As shown, a resource management method 300 according to an embodiment of the present disclosure includes operations S310 to S330.

[0165] In operation S310, a portrait feature of the application running status data is determined based on the application indicator data.

[0166] The application indicator data is obtained by statistically analyzing the application running status data using preset indicators, and the portrait features include at least one of an application granularity portrait, a node granularity portrait, and a cluster granularity portrait.

[0167] The running of an application consumes computer resources, and the application also generates status data based on the consumption and dependency of computer resources. This status data is the application running status data.

[0168] The application running status data may be stored in a database, for example.

[0169] The resource management method of the embodiment of the present disclosure uses preset indicators to obtain application indicator data by statistics of application running status data, and the portrait features of the application running status data determined based on the application indicator data can be understood as a process of mining features that are more representative of application operation from the application status data.

[0170] The preset indicators can be understood as parameter indicators used to evaluate the operation of the application. The application indicator data obtained by using the preset indicators to collect statistics on the application operation status data can be used as the parameter value for evaluating the operation of the application. Therefore, the portrait features of the application operation status data determined based on the application indicator data can characterize the characteristics of the application operation.

[0171] The application granularity portrait, node granularity portrait and cluster granularity portrait represent the portrait features of application granularity, node granularity and cluster granularity respectively.

[0172] The cluster includes at least one node, and each node can run at least one application. Computer resources can be allocated to application granularity, node granularity, and cluster granularity. Therefore, the corresponding profile feature can be determined from at least one of the application granularity, node granularity, and cluster granularity and used as the basis for the subsequent determination of the linkage strategy.

[0173] In operation S320, a linkage strategy is determined according to the portrait features.

[0174] The linkage strategy represents a strategy for linking at least two of application vertical elasticity, application horizontal elasticity, and node elasticity.

[0175] Application vertical elasticity, application horizontal elasticity, and node elasticity have been described in detail above and will not be repeated here.

[0176] In operation S330, resource management is performed according to the linkage strategy.

[0177] In some implementations, application vertical elasticity, application horizontal elasticity, and node elasticity are independent, so the resources required for application operation can be matched with the resources that can be provided by one of application vertical elasticity, application horizontal elasticity, and node elasticity.

[0178] According to the resource management method of the embodiment of the present disclosure, multiple applications of vertical elasticity, horizontal elasticity, and node elasticity can be linked to perform resource management. In the case of linking the above-mentioned multiple independent elasticity strategies, the resource management method of the embodiment of the present disclosure, because the application operation consumes computer resources, uses the image characteristics of the application operation status data as the basis for determining the linkage strategy, which can more accurately match the function of the computer resources. The linkage strategy determined according to the image characteristics can unify the linkage of multiple applications of vertical elasticity, horizontal elasticity, and node elasticity to cope with fluctuations in business demand, especially the "business tide" phenomenon in which the business has a regular fluctuation pattern, and also has higher resource management efficiency. Higher efficiency of resource management is reflected, for example, in the case of linking multiple independent elasticity strategies, the resource management methods are richer, the resource management is more reasonable, and there will be no conflicts in elasticity strategies.

[0179] Figure 4 The schematic diagram schematically shows the resource allocation to the resource deployment unit Pod according to the resource management method of the embodiment of the present disclosure.

[0180] Figure 4 A resource management device that can execute the resource management method of the embodiment of the present disclosure is schematically shown. The resource management device determines the portrait features of the application running status data based on the application indicator data, for example. The application indicator data is obtained by statistically analyzing the application running status data using preset indicators. The portrait features include at least one of the application granularity portrait, the node granularity portrait, and the cluster granularity portrait. According to the portrait features, a linkage strategy is determined. According to the linkage strategy, resource management is performed.

[0181] exist Figure 4 In the example, taking the allocation of resources to the resource deployment unit Pod as an example, a total of x Pods from Pod-1 to Pod-x represent the horizontal elasticity of the application (that is, the number of Pods can be adjusted), and x is an integer greater than or equal to 2. Pod-i is adjusted to Pod-i' to represent the vertical elasticity of the application (that is, the resource specifications of the Pod can be adjusted, and the resource specifications can be understood as the maximum amount of resources used). A total of y nodes from node 1 to node Ny represent the node elasticity (that is, the number of nodes can be adjusted), and y is an integer greater than or equal to 2. According to the resource management method of the embodiment of the present disclosure, at least two of application vertical elasticity, application horizontal elasticity and node elasticity can be linked.

[0182] Preset indicators can be understood as parameter indicators used to evaluate application operation.

[0183] Exemplarily, the preset index may include at least one of a storage resource occupancy index, a computing resource occupancy index, an event transmission rate index, a traffic processing rate index, an operation sensitivity coefficient index, a storage sensitivity coefficient index, and a delay sensitivity coefficient index. The storage resource occupancy index, the computing resource occupancy index, the event transmission rate index, and the traffic processing rate index may be understood as application running state indicators, and the operation sensitivity coefficient index, the storage sensitivity coefficient index, and the delay sensitivity coefficient index may be understood as application characteristic indicators.

[0184] The application running status data includes at least one of running indicator data and application characteristic data.

[0185] The operation indicator data includes at least one of the following: storage resource occupancy data, computing resource occupancy data, event transmission rate, and traffic processing rate.

[0186] The storage resource occupancy data represents the storage resources required for the application to run, and the storage resources include, for example, memory. The computing power resource occupancy data represents the computing power resources required for the application to run, and the computing power resources include, for example, CPU computing power resources and GPU computing power resources.

[0187] The event transmission rate can be represented by, for example, TPS (Transactions Per Second).

[0188] The traffic processing rate may be represented by, for example, QPS (query per second).

[0189] The application characteristic data includes at least one of the following: an operation sensitivity coefficient, a storage sensitivity coefficient, and a delay sensitivity coefficient.

[0190] The computing sensitivity coefficient can be understood as the change in computing power caused by a one-unit change in computing power resources. The storage sensitivity coefficient can be understood as the change in storage capacity caused by a one-unit change in storage resources. The delay sensitivity coefficient can be understood as the change in delay duration caused by a one-unit change in computing power resources.

[0191] The computer resources required for application operation are used for data storage and calculation, etc. According to the resource management method of the embodiment of the present disclosure, in addition to meeting the demand for computer resources to ensure normal business processing, computer resource management can also evaluate the resource management scheme on this basis to select a better and more efficient resource management scheme. For example, operation indicator data such as storage resource occupancy data and computing power resource occupancy data can be used to evaluate the demand for computer resources for normal business processing. For example, application feature data can be used to evaluate the efficiency of computer resource use. For example, a higher computing sensitivity coefficient indicates that the gain of increasing unit computing power resources is higher, so the resource utilization rate is higher, and resource management is more reasonable and efficient.

[0192] Exemplarily, the preset indicators may support maintenance, that is, the user may add, delete, and modify the preset indicators to adapt to different application scenarios, so that the resource management method of the embodiment of the present disclosure may be applied more flexibly.

[0193] According to a resource management method of another embodiment of the present disclosure, for example, the following embodiment can be used to implement a specific example of resource management according to a linkage strategy: according to the linkage strategy, adjust at least one of the following to perform resource management: the number of nodes, the specifications of the resource deployment unit, and the number of resource deployment units.

[0194] The specification of the resource deployment unit, the number of resource deployment units, and each node are related to resources.

[0195] The resource deployment unit is deployed on a node, and the cluster includes at least one node.

[0196] Adjusting the number of nodes is called node elasticity, adjusting the specifications of resource deployment units is called application vertical elasticity, and adjusting the number of resource deployment units is called application horizontal elasticity. In this way, multiple of application vertical elasticity, application horizontal elasticity, and node elasticity can be linked to perform resource management.

[0197] According to a resource management method according to another embodiment of the present disclosure, for example, the following embodiment can be used to implement a specific example of determining a linkage strategy according to portrait features: determining a linkage strategy that meets a target rule according to the portrait features.

[0198] The target rule includes at least one of a service guarantee rule and a resource consumption minimum rule.

[0199] There is the following situation: for a certain portrait feature, multiple linkage strategies can be determined. In this case, according to the resource management method of the embodiment of the present disclosure, determining the linkage strategy that meets the target rule can be understood as a process of screening multiple linkage strategies through the target rule, so that the linkage strategy meets the target rule and resource management is performed according to the linkage strategy that meets the target rule.

[0200] The business assurance rule indicates that the linkage strategy ensures that the normal operation of business applications is less affected. In some cases, resource management will affect the normal operation of business applications. For example, when a new Pod is created and part of the services are transferred to the new Pod, this part of the services may be disconnected for a short time, which is perceptible to the user, so it will affect the user experience. The resource management method of the embodiment of the present disclosure can ensure that the normal operation of business applications is less affected by the linkage strategy that meets the business assurance rule, thereby ensuring the user experience.

[0201] The linkage strategy that satisfies the rule of minimum resource consumption indicates that the linkage strategy has the minimum resource consumption among the optional linkage strategies, thereby improving resource utilization efficiency.

[0202] In the case where the target rule includes a service assurance rule and a resource consumption minimization rule, the priority of the service assurance rule is higher than the priority of the resource consumption minimization rule.

[0203] According to the resource management method of the embodiment of the present disclosure, by determining the linkage strategy that meets the target rules based on the portrait characteristics, it can at least ensure that the normal operation of the business application is less affected or the resource consumption is less, so as to protect the user experience or improve the resource utilization efficiency.

[0204] According to another embodiment of the resource management method of the present disclosure, the portrait feature includes an application granularity portrait. For example, the following embodiments can be used to implement a specific example of determining a linkage strategy based on the portrait feature: determining a strategy for linking application vertical elasticity with application horizontal elasticity based on the application granularity portrait.

[0205] The application granularity portrait is a portrait feature of the application running status data measured in the application dimension. Application vertical elasticity and application horizontal elasticity are also two different elasticity strategies in the application dimension. Compared with node elasticity, the cost of application vertical elasticity and application horizontal elasticity is lower. Therefore, according to the resource management method of the embodiment of the present disclosure, based on the application granularity portrait in the application dimension, a strategy for linking application vertical elasticity and application horizontal elasticity in the application dimension can be determined to achieve unified elastic linkage in the application dimension, which can improve resource management efficiency at low cost.

[0206] According to another embodiment of the resource management method of the present disclosure, the portrait feature includes a node granularity portrait. For example, the following embodiments can be used to implement a specific example of determining a linkage strategy based on the portrait feature: determining a strategy for linking application vertical elasticity with application horizontal elasticity based on application granularity portrait and node granularity portrait.

[0207] In some cases, application vertical elasticity and application horizontal elasticity require that node resources meet conditions. For example, if application horizontal elasticity requires increasing the number of Pods or application vertical elasticity requires increasing the resource specifications of Pods, the current node's resources must be able to accommodate the increased number of Pods or the current node's resources must meet the specifications of the added Pods.

[0208] According to the resource management method of the embodiment of the present disclosure, the node granularity portrait is also used as the basis for determining the linkage strategy between application vertical elasticity and application horizontal elasticity, which can cope with the above situation and more accurately determine the reasonable linkage strategy between application vertical elasticity and application horizontal elasticity.

[0209] Figure 5 A schematic diagram is shown schematically of a resource management method according to another embodiment of the present disclosure, which determines a strategy for linking application vertical elasticity with application horizontal elasticity based on application granularity portraits and node granularity portraits.

[0210] exist Figure 5 In the example of , a specific example of using preset indicators to count application running status data is also shown. The preset indicators can be input into the data configurator, and the data configurator outputs application indicator data. The preset indicators include two types, such as application running status indicators and application feature indicators.

[0211] Figure 5 A specific example of managing portrait features by the portrait manager is also schematically shown ( Figure 5 In the example of , the profile features include application-level profiles and node-level profiles. Figure 5 In the example, the linkage policy represents the strategy of linking application horizontal elasticity and application vertical elasticity. For example, the number of Pods is adjusted to a total of x, and the resource specifications of Pod-1 are also adjusted.

[0212] According to another embodiment of the resource management method of the present disclosure, the portrait feature includes a cluster granularity portrait. For example, the following embodiments can be used to implement a specific example of determining the linkage strategy based on the portrait feature: determining the node elasticity and application vertical elasticity and / or application horizontal elasticity linkage strategy based on the application granularity portrait and the cluster granularity portrait and / or the node granularity portrait.

[0213] Based on the application granularity profile, cluster granularity profile, and / or node granularity profile, the strategies for determining node elasticity and application vertical elasticity and / or application horizontal elasticity linkage include the following situations:

[0214] 1) Determine the node elasticity and application vertical elastic linkage strategies based on the application granularity profile and cluster granularity profile.

[0215] 2) Determine the strategy for node elasticity and application-level elastic linkage based on the application-level and cluster-level portraits.

[0216] 3) Based on the application granularity profile and cluster granularity profile, determine the strategy for node elasticity, application vertical elasticity, and application horizontal elasticity linkage.

[0217] 4) Determine the strategy for node elasticity and application vertical elastic linkage based on the application granularity portrait and node granularity portrait.

[0218] 5) Based on the application granularity portrait and node granularity portrait, determine the strategy for node elasticity and application level elastic linkage.

[0219] 6) Based on the application granularity portrait and node granularity portrait, determine the strategy for node elasticity, application vertical elasticity, and application horizontal elasticity linkage.

[0220] 7) Determine the strategy for node elasticity and application vertical elastic linkage based on the application granularity profile, cluster granularity profile, and node granularity profile.

[0221] 8) Based on the application granularity profile, cluster granularity profile, and node granularity profile, determine the strategy for node elasticity and application level elastic linkage.

[0222] 9) Based on the application granularity profile, cluster granularity profile, and node granularity profile, determine the strategy for node elasticity, application vertical elasticity, and application horizontal elasticity linkage.

[0223] The node granularity portrait is a portrait feature of the application running status data measured from the node dimension, and the cluster granularity portrait is a portrait feature of the application running status data measured from the cluster dimension. The cluster includes at least one node. Therefore, the linkage strategy including the node elasticity strategy can be determined at least based on the node granularity portrait and / or the cluster granularity portrait, thereby unifying and linking multiple independent elasticity strategies to manage resources more efficiently.

[0224] Figure 6 A schematic diagram shows a resource management method according to another embodiment of the present disclosure, which determines the strategy of node elasticity and application vertical elasticity and / or application horizontal elasticity linkage based on application granularity portraits, cluster granularity portraits and / or node granularity portraits.

[0225] and Figure 5 The embodiment shown is different. Figure 6 It also schematically shows an example of a cluster-granularity profile, where the linkage strategy can characterize linkage node elasticity, application vertical elasticity, and / or application horizontal elasticity. Figure 6 The linkage policy shown may also characterize at least two of linkage node elasticity, application vertical elasticity, and application horizontal elasticity.

[0226] Figure 7 The diagram schematically shows a resource management method according to another embodiment of the present disclosure when the application running status data is stored in a database.

[0227] Figure 7 The example in which the database is Prometheus is schematically shown. Prometheus is an open source service monitoring system and time series database.

[0228] Figure 5-Figure 7 In the example, the resource management device is integrated in the application orchestration system Kubernetes as an example, and the resource management device is integrated on any node in the K8S cluster.

[0229] According to a resource management method according to another embodiment of the present disclosure, for example, the following embodiment can be used to implement a specific example of determining a linkage strategy based on portrait features: using a model to process the portrait features to obtain a linkage strategy.

[0230] A model can be understood as a function that receives data as input and generates output. The model may include a machine learning model, for example. The following description will take the model being a machine learning model as an example.

[0231] For example, the model can be pre-trained, the input of the model is the image feature, and the output of the model is the linkage strategy. Model training adjusts the model parameters. By adjusting the model parameters, the performance of the model will also change. When the model meets the requirements, the model training is completed. Therefore, the trained model can meet the predetermined requirements in terms of performance such as accuracy. Therefore, according to the resource management method of the embodiment of the present disclosure, the linkage strategy output by the model at least meets the performance requirements such as accuracy requirements, which is a more intelligent and efficient way to determine the linkage strategy.

[0232] According to a resource management method of another embodiment of the present disclosure, for example, the following embodiments can be used to implement a specific example of using a model to process a portrait feature to obtain a linkage strategy: using at least one first model to process the portrait feature to obtain at least one initial linkage strategy corresponding to each first model. According to the at least one initial linkage strategy, a target linkage strategy is determined.

[0233] When there are plural first models, model parameters of any plural first models are different from each other.

[0234] Exemplarily, for example, multiple first models may be used to process the portrait features to obtain multiple initial linkage strategies corresponding to each first model, and then a target linkage strategy may be determined based on the multiple initial linkage strategies.

[0235] Exemplarily, for example, a first model may be used to process the portrait features to obtain an initial linkage strategy corresponding to the first model. At this time, the initial linkage strategy may be determined as the target linkage strategy.

[0236] The first model is obtained by training with training samples. The model parameters of the first model are different before and after training, and the performance of the first models with different model parameters is also different. The portrait features are processed by multiple first models to obtain multiple initial linkage strategies and determine the target linkage strategy based on the multiple initial linkage strategies. The advantages of each first model can be combined to avoid inaccurate linkage strategies caused by the defects of a certain first model. Therefore, the resource management method according to the embodiment of the present disclosure can determine a more accurate and reasonable linkage strategy, thereby correspondingly improving the resource management efficiency.

[0237] Exemplarily, in the case where the initial linkage strategies include multiple ones, a target linkage strategy may be determined from the multiple initial linkage strategies based on a target rule, for example.

[0238] Exemplarily, when the first model includes a plurality of models, each first model may be stored in a model library, for example. For example, the model library may support calling through an API (Application Programming Interface, referred to as interface).

[0239] exist Figure 5-Figure 7 In the example, it is schematically shown that the first model includes multiple specific examples in which each first model is stored in the database, and the multiple first models include, for example, at least a first model M1-1 and a first model M1-2.

[0240] Exemplarily, the first model may also support maintenance, for example, the user may add, delete, and update the first model.

[0241] Exemplarily, according to a resource management method of another embodiment of the present disclosure, when there are multiple first models, the multiple first models are trained based on training samples of different business scenarios so that the model parameters of each first model are different.

[0242] The first model is trained based on the training samples, and the first models trained by the training samples of different business scenarios are different. For example, the first model trained by the training samples of a certain business scenario better learns the characteristics of this business scenario, so the performance of the first model is better for the input data of the business scenario. If the first model is not trained by the training samples of another business scenario, the performance of the first model for the input data of the other business scenario is poor.

[0243] Due to the diversity and complexity of business scenarios, it is difficult to cope with all business scenarios through a certain first model and to have excellent performance in all business scenarios. Therefore, according to the resource management method of the embodiment of the present invention, multiple first models obtained by training with training samples of different business scenarios can distinguish different application scenarios to accurately determine the linkage strategy, thereby improving resource management efficiency.

[0244] Exemplarily, according to another embodiment of the resource management method of the present disclosure, for example, it can also support the user to select at least one specific first model, and determine the target linkage strategy according to the initial linkage strategy output by the selected first model.

[0245] Exemplarily, according to a resource management method of another embodiment of the present disclosure, for example, input data such as at least portrait features can be analyzed through preset logic to obtain a business scenario matching the portrait features, and an initial linkage strategy can be determined based on a first model of the matching business scenario.

[0246] According to another embodiment of the resource management method of the present disclosure, for example, the following embodiment can be used to implement a specific example of determining a target linkage strategy based on at least one initial linkage strategy: using a second model to process at least one initial linkage strategy and portrait features and / or application indicator data to obtain a target linkage strategy.

[0247] In the case where there are multiple initial linkage strategies, it is necessary to determine a target linkage strategy from the multiple initial linkage strategies. This process can be understood as analyzing multiple candidate initial linkage strategies to determine the target linkage strategy.

[0248] When it is necessary to comprehensively analyze and compare each initial linkage strategy to determine the target linkage strategy, it is relatively simple to use only the initial linkage strategy as a basis. Therefore, on this basis, the target linkage strategy can be comprehensively and accurately determined based on at least one initial linkage strategy and portrait features and / or application indicator data.

[0249] Exemplarily, the second model may also be a trained machine learning model, thereby determining the target linkage strategy more intelligently and efficiently.

[0250] Exemplarily, the second model may also support maintenance, for example, supporting the user to update the second model.

[0251] Exemplarily, the second model may also be stored in a model library, and the second model may be used by calling an API of the model library.

[0252] exist Figure 5-Figure 7 In the example of FIG. 1 , a specific example of the second model M2 is schematically shown.

[0253] According to a resource management method of another embodiment of the present disclosure, for example, the following embodiments can be used to implement a specific example of determining a linkage strategy based on a portrait feature: Determine application indicator data at a future time based on the portrait feature. Determine the required specifications and required quantity of a resource deployment unit based on the application indicator data at a future time. Determine a linkage strategy based on the required specifications and required quantity of a resource deployment unit.

[0254] When the application is running, the application indicator data is dependent in the time dimension. For example, the Long Short-Term Memory (LSTM) network can be used to determine the application indicator data at the future moment based on the image features. The LSTM network is a time recursive neural network with long-term memory capabilities. The network structure of the LSTM network includes one or more neurons with forgettable and memory functions, which can be used to selectively "memorize" the output results of the previous neuron. Therefore, the LSTM network can adapt to the characteristics of the application indicator data being dependent in the time dimension, thereby accurately predicting the application indicator data at the future moment.

[0255] A resource deployment unit can be understood as a unitized resource deployment instance. Each resource deployment unit can be deployed independently, and the resources and environment of each resource deployment unit are shared. A resource deployment unit can include a Pod, for example. The following example will take the resource deployment unit as a Pod.

[0256] The portrait features of the application running status data represent the characteristics of the application running. Based on the portrait features, the application index data at the future moment can be predicted. Based on the application index data at the future moment, the demand specifications and demand quantities of the resource deployment unit can be accurately determined, and then a reasonable and efficient linkage strategy can be determined. The demand specifications and demand quantities of the resource deployment unit can be understood as the resource demand at the granularity of the resource deployment unit that matches the future application index data. The resource demand is represented by the demand specifications and demand quantities of the resource deployment unit.

[0257] Adjusting the demand specifications of the resource deployment unit can be understood as applying vertical elasticity, and adjusting the demand quantity of resource deployment unit instances can be understood as applying horizontal elasticity.

[0258] For example, the maximum flow minimum cut method can be used to determine the demand specifications and the required quantity of the resource deployment unit according to the application indicator data at the future moment. The maximum flow minimum cut method means that in a network flow, the maximum flow that can reach the sink from the source point is equal to the minimum capacity and the set of edges that can cause the network flow to be interrupted if removed from the network. That is, in any network, the value of the maximum flow is equal to the capacity of the minimum cut. The maximum flow minimum cut method can reasonably allocate resources and improve the overall efficiency of the network system.

[0259] According to a resource management method of another embodiment of the present disclosure, for example, the following embodiments can be used to implement a specific example of determining the required specifications and required quantities of resource deployment units according to application indicator data at a future moment: Determine a first specification and a first quantity of resource deployment units that match the application indicator data at a historical moment. Determine a second specification and a second quantity of resource deployment units that match the application indicator data at a future moment. Determine the required specifications and required quantity of the resource deployment unit based on the first specification and the first quantity and the second specification and the second quantity.

[0260] The application indicator data at the historical moment is obtained by statistically analyzing the application running status data at the historical moment using preset indicators.

[0261] Exemplarily, for example, based on the application indicator data at the historical moment, the first specification and the first quantity of the resource deployment units matching the application indicator data at the historical moment may be determined.

[0262] The operation of a business application at any historical moment corresponds to a real resource usage data, so the first specification and the first quantity of the resource deployment unit matching the application indicator data at the historical moment can be determined according to the application indicator data at the historical moment. For example, the application indicator data at the historical moment can be mapped to the first specification and the first quantity of the resource deployment unit matching it.

[0263] For example, the predicted application indicator data at the future moment can be compared with the application indicator data at the historical moment, and the application indicator data with the same value or the closest value can be determined. The second specification and the second quantity of the resource deployment unit that matches the application indicator data at the future moment can also be determined based on the mapping relationship between the application indicator data and the first specification and the first quantity. That is, for example, the application indicator data at the future moment can be numerically matched with the application indicator data at the historical moment. When the values ​​of the two are the same or close, the second specification and the second quantity can be taken to be the same as the first specification and the first quantity.

[0264] The following situation exists: the first specification and the first quantity corresponding to the application indicator data at the historical moment indicate the need for expansion, while the second specification and the second quantity corresponding to the application indicator data at the future moment indicate the need for reduction. In this case, the resources released by the future reduction can offset at least part of the resources required for the historical expansion. The offset part can be determined based on the first specification and the first quantity and the second specification and the second quantity. The resources that actually need to be allocated are, for example, the required specifications and the required quantities of the resource deployment unit.

[0265] According to the resource management method of the embodiment of the present disclosure, based on the dependency characteristics of resource management in the time dimension, the complementary resource demands of historical moments and future moments are offset, so that a more reasonable resource management plan can be determined and the resource management efficiency can be improved.

[0266] For example, in order to more closely connect the resource requirements of historical moments and future moments in the time dimension to improve resource management efficiency, the span between historical moments and future moments can be set to be smaller. For example, the historical moment can be the current moment, and the future moment can be the next moment of the current moment.

[0267] According to a resource management method of another embodiment of the present disclosure, for example, the following embodiments can be used to implement a specific example of determining the image features of the application running status data according to the application indicator data: according to the application indicator data, determine at least one of the relationship between the application indicator data and the resources, the topological relationship between the resource deployment units, and the available resources of the node. According to at least one of the relationship between the application indicator data and the resources, the topological relationship between the resource deployment units, and the available resources of the node, determine the image features.

[0268] Exemplarily, for example, the relationship between application indicator data and resources, the topological relationship between resource deployment units, and at least one of the available resources of a node may be used as a portrait feature.

[0269] The profile features of the application running status data represent the features of the application running, and the profile features of the application running status data are subsequently used to allocate resources. The relationship between the application indicator data and resources can represent the correlation between resource usage caused by the application running, so it can be used as a dimension to evaluate the features of the application running (i.e., the profile features), and can be used to assist in resource management later.

[0270] Exemplarily, the relationship between application indicator data and resources can be represented by the relationship between QPS and resources, for example.

[0271] Since the resource management method of the embodiment of the present disclosure can perform resource management at the granularity of resource deployment units, for example, resource management can be performed by adjusting the resource specifications and quantity of resource deployment units, and the topological relationship between resource deployment units can characterize the overall relationship of resource deployment units, the topological relationship between resource deployment units can be used as the basis for resource management, making the resource management of the embodiment of the present disclosure more reasonable and efficient.

[0272] Exemplarily, the topological relationship between the resource deployment units can be obtained by constructing an adjacency matrix based on the calling relationship between the resource deployment units. For example, a topological graph of the service can also be constructed based on the adjacency matrix.

[0273] The available resources of a node can be understood as the resources of a node that can be used for resource management. The resource management method of the disclosed embodiment, when performing resource management through the application dimension, still needs to be based on the premise that the node resources meet the conditions. In some cases, a part of the node resources is used to maintain the basic functions of the node, etc. If this part of the resources is used for allocation, it will affect the basic functions. At this time, the available resources of the node can be obtained by the difference between the total resources of the node and the resources required to maintain the basic functions. The available resources of the node can accurately represent the resources that can actually be used for resource management. For example, accurate and efficient resource management can be performed based on the available resources of the node in the future.

[0274] According to a resource management method of another embodiment of the present disclosure, for example, the following embodiments can be used to implement a specific example of determining the required specifications and required quantities of resource deployment units according to application indicator data at a future moment: determining a target resource quantity that matches the application indicator data at a future moment. Determine the required specifications and required quantities of resource deployment units according to the target resource quantity.

[0275] The target resource volume is determined based on the relationship between application indicator data and resources.

[0276] Exemplarily, the target resource quantity may be represented by the required specification and required quantity of the resource deployment unit. The relationship between the application indicator data and the resource may represent the relationship between the application indicator data and the specification and / or quantity of the resource deployment unit.

[0277] Exemplarily, the target resource quantity can also be represented by computer resources, for example, and the relationship between application indicator data and resources can, for example, represent the relationship between application indicator data and computer resources. In this case, for example, the mapping relationship between computer resources and the specifications and / or quantities of resource deployment units can also be determined in advance, so that the required specifications and required quantities of resource deployment units can be determined according to the target resource quantity.

[0278] It should also be noted that the required specifications and required quantity of resource deployment units can be understood as the specifications and quantity of resource deployment units required to process application indicator data at future times (i.e., the total number of resource deployment units), or the required specifications and required quantity of resource deployment units can also be understood as the adjustment amount of the specifications and quantity of resource deployment units required to process application indicator data at future times.

[0279] The resource management method of the disclosed embodiment utilizes the service provided by the resource deployment unit to process application indicator data: the target resource quantity that matches the application indicator data at the future moment can be determined based on the relationship between the application indicator data and the resources, and the target resource quantity can be allocated to the granularity of the resource deployment unit, that is, the required specifications and required quantity of the resource deployment unit are determined based on the target resource quantity. For example, the specifications and quantity of the resource deployment unit can be adjusted based on the required specifications and required quantity of the resource deployment unit, so that the resource deployment unit after resource management matches the application indicator data at the future moment. Therefore, the resource deployment unit after resource management has the ability to cope with the application indicator data at the future moment.

[0280] According to a resource management method according to another embodiment of the present disclosure, for example, the following embodiments can be used to implement a specific example of determining a target resource quantity that matches application indicator data at a future moment: according to the relationship between the application indicator data and the resource, at least one resource quantity data that matches the application indicator data at a future moment is determined. A target resource quantity with the best cost-effectiveness is determined from the at least one resource quantity data.

[0281] The best price / performance ratio is determined based on the minimum value of the ratio between resources and application indicator data.

[0282] In the case where the relationship between application indicator data and resources is one-to-many, multiple resource quantity data can be determined based on the application indicator data of a certain value at a future moment. According to the resource management method of the embodiment of the present disclosure, the target resource quantity can be determined from multiple resource quantity data by the parameter of the best cost performance ratio. The ratio between resources and application indicator data can characterize the resources required for each processing of an application indicator data. The minimum value of the ratio between resources and application indicator data indicates that the resources required for each processing of an application indicator data are the minimum, that is, the resource utilization rate is higher. Therefore, the resource utilization rate corresponding to the target resource quantity determined by the resource management method of the embodiment of the present disclosure is higher, and the resource management efficiency is also higher.

[0283] According to another embodiment of the resource management method of the present disclosure, for example, the following embodiment can be used to implement a specific example of adjusting the specifications of the resource deployment unit and / or the number of resource deployment units for resource management: calling the resources of the target node in the candidate node to adjust the specifications of the resource deployment unit and / or the number of resource deployment units.

[0284] The resources of the target node meet the resource requirements of the specification of the resource deployment unit and / or the number of resource deployment units, and the remaining resources of the target node after the resources are called are the smallest among the candidate nodes.

[0285] In the case of calling node resources for application-dimensional elasticity, there may be multiple candidate nodes available for resource scheduling. According to the resource management method of the embodiment of the present disclosure, since the resourceization of the target node meets the resource requirements of the specifications of the resource deployment unit and / or the number of resource deployment units, and the remaining resources of the target node after being called are the smallest among the candidate nodes, this indicates that the target node can also achieve the maximum node resource utilization efficiency under the premise of meeting the resource requirements of the specifications of the resource deployment unit and / or the number of resource deployment units. Therefore, resource management is more efficient by calling the resources of the target node in the candidate node to adjust the specifications of the resource deployment unit and / or the number of resource deployment units, and the resource utilization rate is also higher.

[0286] According to a resource management method according to another embodiment of the present disclosure, for example, the following embodiments may be used to implement a specific example of performing resource management according to a linkage strategy: determining a linkage strategy instruction according to the linkage strategy.

[0287] The linkage policy instruction is used to instruct the execution of the linkage policy to complete resource management.

[0288] Exemplarily, for example, the executor may respond to the received linkage strategy instruction and execute the linkage strategy to complete resource management.

[0289] According to the resource management method of the embodiment of the present disclosure, the determination of the linkage strategy and the execution of the linkage strategy can be decoupled. After the linkage strategy is determined, the linkage strategy instruction instructs the execution of the linkage strategy. For example, it supports separately updating the determination process of the linkage strategy and the execution process of the linkage strategy, which is more flexible.

[0290] Figure 8 A schematic diagram of a resource management method according to another embodiment of the present disclosure is schematically shown.

[0291] and Figure 5-Figure 7 The embodiments are different. Figure 8 The resource management device for executing the resource management method is shown as a cloud service. The cloud service of the resource management device can be deployed in a cloud native ecosystem, and provide elasticity strategies and resource management services for other cloud native applications.

[0292] like Figure 8 As shown, according to the resource management method of the embodiment of the present disclosure, the user can, for example, configure in advance the preset indicators required to determine at least one of the application granularity portrait, the node granularity portrait, and the cluster granularity portrait by calling the cloud service interface. The cloud service interface can also support the maintenance of the first model, the second model, etc.

[0293] exist Figure 5-Figure 8In the example, a specific example of triggering the execution of the elastic policy through an elastic trigger is also schematically shown. The execution of the elastic policy can be performed by the corresponding executor, for example, application vertical elasticity is executed by the application vertical elasticity executor, application horizontal elasticity is executed by the application horizontal elasticity executor, and node elasticity is executed by the node elasticity executor.

[0294] For example, elastic triggers and actuators can be built-in or external to the K8S cluster.

[0295] Compared to Figure 7 The embodiment of deploying a resource management device on a node of a K8S cluster is shown. Figure 8 The resource management device shown does not need to be uniformly deployed as a cloud service, and only elastic actuators need to be deployed.

[0296] In summary, according to the resource management method of the embodiment of the present disclosure, users can be provided with at least two linked strategies among application horizontal elasticity, application vertical elasticity and node elasticity, thereby solving the problem of elasticity conflicts that may arise when application horizontal elasticity, application vertical elasticity and node elasticity are each an independent elasticity strategy.

[0297] In addition, the portrait features determined based on application indicator data are more representative, and thus more reasonable linkage strategies can be determined to meet business needs and improve resource utilization.

[0298] Furthermore, according to the resource management method of the embodiment of the present invention, the application indicator data at the future moment can be accurately predicted based on the application indicator data at the historical moment, and the demand specifications and demand quantity of the resource deployment units that match the application indicator data at the future moment can be determined, so that resources can be prepared in advance and the probability of vicious expansion caused by business surges can be reduced.

[0299] Fig. 9 The block diagram schematically shows a resource management device according to an embodiment of the present disclosure.

[0300] like Fig. 9 As shown, the resource management device 900 of the embodiment of the present disclosure includes, for example: a portrait feature determination module 910, a linkage strategy determination module 920 and a resource management module 930.

[0301] The image feature determination module 910 is used to determine the image features of the application running status data according to the application indicator data.

[0302] The application indicator data is obtained by statistically analyzing the application running status data using preset indicators, and the portrait features include at least one of an application granularity portrait, a node granularity portrait, and a cluster granularity portrait.

[0303] The linkage strategy determination module 920 is used to determine the linkage strategy according to the portrait features.

[0304] The linkage strategy represents a strategy for linking at least two of application vertical elasticity, application horizontal elasticity, and node elasticity.

[0305] The resource management module 930 is used to perform resource management according to the linkage strategy.

[0306] According to a resource management device of another embodiment of the present disclosure, the linkage strategy determination module includes: a linkage strategy first determination submodule, which is used to determine a linkage strategy that meets the target rule based on the portrait characteristics, wherein the target rule includes at least one of a business assurance rule and a minimum resource consumption rule, wherein, when the target rule includes the business assurance rule and the minimum resource consumption rule, the priority of the business assurance rule is higher than the priority of the minimum resource consumption rule.

[0307] According to a resource management device of another embodiment of the present disclosure, in which the portrait feature includes an application granularity portrait, the linkage strategy determination module includes: a linkage strategy second determination submodule, which is used to determine the strategy of linking application vertical elasticity and application horizontal elasticity based on the application granularity portrait.

[0308] According to a resource management device of another embodiment of the present disclosure, in which the portrait feature includes a node granularity portrait, the linkage strategy determination module includes: a linkage strategy third determination submodule, which is used to determine the strategy of linking application vertical elasticity and application horizontal elasticity based on the application granularity portrait and the node granularity portrait.

[0309] According to a resource management device of another embodiment of the present disclosure, in which, when the portrait feature includes a cluster granularity portrait, the linkage strategy determination module includes: a fourth linkage strategy determination submodule, used to determine the node elasticity and application vertical elasticity and / or application horizontal elasticity linkage strategy based on the application granularity portrait and the cluster granularity portrait and / or the node granularity portrait.

[0310] According to a resource management device of another embodiment of the present disclosure, the linkage strategy determination module includes: a linkage strategy fifth determination submodule, which is used to process the image features using a model to obtain a linkage strategy.

[0311] According to a resource management device according to yet another embodiment of the present disclosure, a fifth linkage strategy determination submodule includes: an initial linkage strategy determination unit and a target linkage strategy determination unit.

[0312] The initial linkage strategy determination unit is used to process the portrait features using at least one first model to obtain at least one initial linkage strategy corresponding to each first model, wherein when there are multiple first models, the model parameters of any multiple first models are different from each other.

[0313] The target linkage strategy determining unit is used to determine a target linkage strategy according to at least one initial linkage strategy.

[0314] According to a resource management device of another embodiment of the present disclosure, when there are multiple first models, the multiple first models are trained based on training samples of different business scenarios so that model parameters of each first model are different.

[0315] According to a resource management device of another embodiment of the present disclosure, the target linkage strategy determination unit includes: a target linkage strategy determination subunit, which is used to use the second model to process at least one initial linkage strategy and portrait features and / or application indicator data to obtain a target linkage strategy.

[0316] According to a resource management device of yet another embodiment of the present disclosure, the linkage strategy determination module includes: a prediction submodule, a demand resource determination submodule, and a linkage strategy sixth determination submodule.

[0317] The prediction submodule is used to determine the application indicator data at future moments based on the portrait features.

[0318] The demand resource determination submodule is used to determine the demand specifications and quantity of resource deployment units based on the application indicator data at a future moment.

[0319] The sixth linkage strategy determination submodule is used to determine the linkage strategy according to the demand specifications and demand quantities of the resource deployment units.

[0320] According to a resource management device according to yet another embodiment of the present disclosure, a required resource determination submodule includes: a first matching unit, a second matching unit, and a required resource determination unit.

[0321] The first matching unit is used to determine a first specification and a first quantity of resource deployment units that match application indicator data at a historical moment, wherein the application indicator data at a historical moment is obtained by statistically analyzing application running status data at a historical moment using preset indicators.

[0322] The second matching unit is used to determine a second specification and a second quantity of resource deployment units that match the application indicator data at a future moment.

[0323] The demand resource determination unit is used to determine the demand specification and the demand quantity of the resource deployment unit according to the first specification and the first quantity and the second specification and the second quantity.

[0324] According to a resource management device of yet another embodiment of the present disclosure, the image feature determination module includes: an image feature dimension determination submodule and an image feature determination submodule.

[0325] The portrait feature dimension determination submodule is used to determine at least one of the relationship between application indicator data and resources, the topological relationship between resource deployment units, and the available resources of the node based on the application indicator data.

[0326] The portrait feature determination submodule is used to determine the portrait feature based on at least one of the relationship between application indicator data and resources, the topological relationship between resource deployment units, and the available resources of the node.

[0327] According to a resource management device according to yet another embodiment of the present disclosure, the required resource determination submodule includes: a target resource quantity determination unit and a required resource determination unit.

[0328] The target resource amount determination unit is used to determine the target resource amount that matches the application indicator data at a future moment, wherein the target resource amount is determined based on the relationship between the application indicator data and the resources.

[0329] The demand resource determination unit is used to determine the demand specifications and demand quantities of the resource deployment unit according to the target resource quantity.

[0330] According to a resource management device according to yet another embodiment of the present disclosure, the target resource amount determination unit includes: a resource matching subunit and a target resource amount determination subunit.

[0331] The resource matching subunit is used to determine at least one resource quantity data matching the application indicator data at a future moment according to the relationship between the application indicator data and the resources.

[0332] The target resource quantity determination subunit is used to determine the target resource quantity with the best cost performance ratio from at least one resource quantity data, wherein the best cost performance ratio is determined based on the minimum value of the ratio between the resource and the application indicator data.

[0333] According to another embodiment of the resource management device of the present disclosure, the resource management module includes: an adjustment submodule, which is used to adjust at least one of the following according to the linkage strategy to perform resource management: the number of nodes, the specifications of the resource deployment units, and the number of resource deployment units, wherein the specifications of the resource deployment units, the number of resource deployment units and each node are related to resources, the resource deployment units are deployed on the nodes, and the cluster includes at least one node.

[0334] According to another embodiment of the resource management device of the present disclosure, the adjustment submodule includes: a node resource calling unit, which is used to call the resources of the target node among the candidate nodes to adjust the specifications of the resource deployment units and / or the number of resource deployment units, wherein the resources of the target node meet the resource requirements of the specifications of the resource deployment units and / or the number of resource deployment units, and the remaining resources of the target node after the resources are called are the smallest among the candidate nodes.

[0335] According to a resource management device of another embodiment of the present disclosure, the resource management module further includes: a linkage policy instruction determination submodule, which is used to determine a linkage policy instruction according to the linkage policy, wherein the linkage policy instruction is used to instruct execution of the linkage policy to complete resource management.

[0336] According to another embodiment of the resource management device of the present disclosure, the application indicator data includes at least one of operation indicator data and application characteristic data, wherein the operation indicator data includes at least one of the following: storage resource occupancy data, computing power resource occupancy data, event transmission rate and traffic processing rate, and the application characteristic data includes at least one of the following: operation sensitivity coefficient, storage sensitivity coefficient and delay sensitivity coefficient.

[0337] It should be understood that the embodiments of the device part of the present disclosure are the same or similar to the embodiments of the method part of the present disclosure, and the technical problems solved and the technical effects achieved are also the same or similar, and the present disclosure will not elaborate on them here.

[0338] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a computer-readable storage medium, and a computer program product.

[0339] Fig.10 A schematic block diagram of an example electronic device 1000 that can be used to implement an embodiment of the present disclosure is shown. Electronic devices include various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic devices may also include various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0340] like Fig.10As shown, the electronic device 1000 includes a computing unit 1001, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 1002 or a computer program loaded from a storage unit 1008 to a random access memory (RAM) 1003. In the RAM 1003, various programs and data required for the operation of the electronic device 1000 can also be stored. The computing unit 1001, the ROM 1002, and the RAM 1003 are connected to each other via a bus 1004. An input / output (I / O) interface 1005 is also connected to the bus 1004.

[0341] Multiple components in the electronic device 1000 are connected to the I / O interface 1005, including: an input unit 1006, such as a keyboard, a mouse, etc.; an output unit 1007, such as various types of displays, speakers, etc.; a storage unit 1008, such as a disk, an optical disk, etc.; and a communication unit 1009, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 1009 allows the device 1000 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0342] The computing unit 1001 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 1001 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 1001 performs the various methods and processes described above, such as the resource management method. For example, in some embodiments, the aforementioned method may be implemented as a computer software program, which is tangibly included in a machine-readable medium, such as a storage unit 1008. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 1000 via the ROM 1002 and / or the communication unit 1009. When the computer program is loaded into the RAM 1003 and executed by the computing unit 1001, one or more steps of the resource management method described above may be performed. Alternatively, in other embodiments, the computing unit 1001 may be configured to execute the resource management method in any other appropriate manner (eg, by means of firmware).

[0343] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems on a chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0344] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0345] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, a flash memory, or any suitable combination of the foregoing.

[0346] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a cathode ray tube (CRT) or a liquid crystal display (LCD) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0347] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0348] A computer system may include clients and servers. Clients and servers are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship to each other.

[0349] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0350] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A resource management method, characterized in that: include: Determine, according to the application indicator data, a portrait feature of the application running status data, wherein the application indicator data is obtained by statistically analyzing the application running status data using preset indicators, and the portrait feature includes at least one of an application granularity portrait, a node granularity portrait, and a cluster granularity portrait; Determine a linkage strategy according to the portrait feature, wherein the linkage strategy represents a strategy for linking at least two of application vertical elasticity, application horizontal elasticity, and node elasticity; Resource management is performed according to the linkage strategy.

2. The method according to claim 1, characterized in that in, Determining the linkage strategy according to the portrait features includes: According to the portrait features, a linkage strategy that satisfies the target rules is determined, wherein the target rules include at least one of business assurance rules and minimum resource consumption rules, and wherein, when the target rules include the business assurance rules and the minimum resource consumption rules, the priority of the business assurance rules is higher than the priority of the minimum resource consumption rules.

3. The method according to claim 1, characterized in that in, When the image feature includes the application granularity image, determining the linkage strategy according to the image feature includes: According to the application granularity profile, a strategy for linking the application vertical elasticity and the application horizontal elasticity is determined.

4. The method according to claim 1, characterized in that in, When the portrait feature includes the node granularity portrait, determining the linkage strategy according to the portrait feature includes: A strategy for linking the application vertical elasticity and the application horizontal elasticity is determined according to the application granularity portrait and the node granularity portrait.

5. The method according to claim 1, characterized in that in, When the image feature includes the cluster granularity image, determining the linkage strategy according to the image feature includes: According to the application granularity portrait, the cluster granularity portrait and / or the node granularity portrait, a strategy for linking the node elasticity and the application vertical elasticity and / or the application horizontal elasticity is determined.

6. The method according to any one of claims 1 to 5, characterized in that Determining the linkage strategy according to the portrait features includes: The image features are processed using a model to obtain the linkage strategy.

7. The method according to claim 6, characterized in that The using model to process the image features to obtain the linkage strategy includes: Processing the image features using at least one first model to obtain at least one initial linkage strategy corresponding to each of the first models, wherein, when there are multiple first models, model parameters of any multiple first models are different from each other; A target linkage strategy is determined according to the at least one initial linkage strategy.

8. The method according to claim 7, characterized in that in, When there are multiple first models, the multiple first models are trained based on training samples of different business scenarios, so that the model parameters of each first model are different.

9. The method according to claim 7, characterized in that: in, The determining of the target linkage strategy according to the at least one initial linkage strategy includes: The at least one initial linkage strategy and the portrait features and / or the application indicator data are processed using a second model to obtain the target linkage strategy.

10. The method according to any one of claims 1 to 5, characterized in that in, Determining the linkage strategy according to the portrait features includes: Determine application indicator data at a future time based on the portrait features; Determine the required specifications and required quantity of resource deployment units according to the application indicator data at the future moment; The linkage strategy is determined according to the demand specifications and the demand quantity of the resource deployment unit.

11. The method according to claim 10, characterized in that in, Determining the required specifications and required quantity of resource deployment units according to the application indicator data at the future moment includes: Determining a first specification and a first quantity of the resource deployment units that match the application indicator data at a historical moment, wherein the application indicator data at a historical moment is obtained by statistically analyzing the application running state data at a historical moment using the preset indicator; Determining a second specification and a second quantity of the resource deployment units that match the application indicator data at the future moment; The required specification and the required quantity of the resource deployment unit are determined according to the first specification, the first quantity and the second specification, the second quantity.

12. The method according to any one of claims 1 to 5, characterized in that in, Determining the image features of the application running status data according to the application indicator data includes: Determine, according to the application indicator data, at least one of a relationship between the application indicator data and resources, a topological relationship between resource deployment units, and available resources of a node; The portrait feature is determined based on at least one of the relationship between the application indicator data and resources, the topological relationship between resource deployment units, and the available resources of the node.

13. The method according to claim 10, characterized in that in, Determining the required specifications and required quantity of resource deployment units according to the application indicator data at the future moment includes: Determining a target resource amount that matches the application indicator data at the future moment, wherein the target resource amount is determined based on a relationship between the application indicator data and resources; According to the target resource quantity, the required specification and required quantity of the resource deployment unit are determined.

14. The method according to claim 13, characterized in that in, The determining of the target resource amount matching the application indicator data at the future moment comprises: Determine, according to the relationship between the application indicator data and the resource, at least one resource quantity data matching the application indicator data at the future moment; The target resource quantity with the best cost performance ratio is determined from the at least one resource quantity data, wherein the best cost performance ratio is determined according to the minimum value of the ratio between the resource and the application indicator data.

15. The method according to any one of claims 1 to 5, characterized in that in, The performing resource management according to the linkage strategy includes: According to the linkage strategy, at least one of the following is adjusted to perform resource management: the number of nodes, the specifications of the resource deployment units, and the number of the resource deployment units, wherein the specifications of the resource deployment units, the number of the resource deployment units, and each of the nodes are related to the resources, the resource deployment units are deployed on the nodes, and the cluster includes at least one of the nodes.

16. The method according to claim 15, characterized in that in, Adjusting the specification of the resource deployment unit and / or the number of the resource deployment units to perform resource management includes: Call the resources of the target node among the candidate nodes to adjust the specifications of the resource deployment unit and / or the number of the resource deployment units, wherein the resources of the target node meet the resource requirements of the specifications of the resource deployment unit and / or the number of the resource deployment units, and the remaining resources of the target node after the resources are called are the smallest among the candidate nodes.

17. The method according to any one of claims 1 to 5, characterized in that in, The performing resource management according to the linkage strategy further includes: According to the linkage strategy, a linkage strategy instruction is determined, wherein the linkage strategy instruction is used to instruct execution of the linkage strategy to complete resource management.

18. The method according to any one of claims 1 to 5, characterized in that in, The application indicator data includes at least one of operation indicator data and application characteristic data, wherein the operation indicator data includes at least one of the following: storage resource occupancy data, computing resource occupancy data, event transmission rate and traffic processing rate; the application characteristic data includes at least one of the following: operation sensitivity coefficient, storage sensitivity coefficient and delay sensitivity coefficient.

19. The method according to claim 9, characterized in that in, At least one of the first model, the second model, and the preset indicator supports maintenance.

20. A resource management device, characterized in that: include: A portrait feature determination module, used to determine the portrait features of the application running status data according to the application indicator data, wherein the application indicator data is obtained by statistically analyzing the application running status data using preset indicators, and the portrait features include at least one of an application granularity portrait, a node granularity portrait, and a cluster granularity portrait; A linkage strategy determination module, used to determine a linkage strategy according to the portrait features, wherein the linkage strategy represents a strategy for linking at least two of application vertical elasticity, application horizontal elasticity, and node elasticity; The resource management module is used to perform resource management according to the linkage strategy.

21. The device according to claim 20, characterized in that in, The linkage strategy determination module includes: The first linkage strategy determination submodule is used to determine the linkage strategy that meets the target rules based on the portrait characteristics, wherein the target rules include at least one of the business assurance rules and the minimum resource consumption rules, and wherein, when the target rules include the business assurance rules and the minimum resource consumption rules, the priority of the business assurance rules is higher than the priority of the minimum resource consumption rules.

22. The device according to claim 20, characterized in that in, When the image feature includes the application granularity image, the linkage strategy determination module includes: The second linkage strategy determination submodule is used to determine the linkage strategy between the vertical elasticity of the application and the horizontal elasticity of the application according to the application granularity profile.

23. The device according to claim 20, characterized in that in, When the portrait feature includes the node granularity portrait, the linkage strategy determination module includes: The third linkage strategy determination submodule is used to determine the linkage strategy of the application vertical elasticity and the application horizontal elasticity according to the application granularity portrait and the node granularity portrait.

24. The device according to claim 20, characterized in that in, When the image feature includes the cluster granularity image, the linkage strategy determination module includes: The fourth linkage strategy determination submodule is used to determine the node elasticity and the application vertical elasticity and / or the application horizontal elasticity linkage strategy according to the application granularity portrait and the cluster granularity portrait and / or the node granularity portrait.

25. The device according to any one of claims 20 to 24, characterized in that The linkage strategy determination module includes: The fifth linkage strategy determination submodule is used to process the image features using a model to obtain the linkage strategy.

26. The device according to claim 25, characterized in that The fifth determination submodule of the linkage strategy includes: an initial linkage strategy determination unit, configured to process the portrait features using at least one first model to obtain at least one initial linkage strategy corresponding to each of the first models, wherein, when there are multiple first models, model parameters of any multiple first models are different from each other; The target linkage strategy determining unit is used to determine a target linkage strategy according to the at least one initial linkage strategy.

27. The device according to claim 26, characterized in that in, When there are multiple first models, the multiple first models are trained based on training samples of different business scenarios, so that the model parameters of each first model are different.

28. The device according to claim 26, characterized in that in, The target linkage strategy determination unit includes: The target linkage strategy determination subunit is used to use the second model to process the at least one initial linkage strategy and the portrait features and / or the application indicator data to obtain the target linkage strategy.

29. The device according to any one of claims 20 to 24, characterized in that in, The linkage strategy determination module includes: A prediction submodule, used to determine application indicator data at a future moment according to the image features; A demand resource determination submodule, used to determine the demand specifications and the required quantity of the resource deployment unit according to the application indicator data at the future moment; The sixth linkage strategy determination submodule is used to determine the linkage strategy according to the demand specifications and the demand quantity of the resource deployment unit.

30. The device according to claim 29, characterized in that in, The demand resource determination submodule includes: A first matching unit, configured to determine a first specification and a first quantity of the resource deployment unit that matches the application indicator data at a historical moment, wherein the application indicator data at the historical moment is obtained by statistically analyzing the application running state data at the historical moment using the preset indicator; A second matching unit, configured to determine a second specification and a second quantity of the resource deployment units that match the application indicator data at the future moment; A required resource determination unit is used to determine the required specification and the required quantity of the resource deployment unit according to the first specification, the first quantity and the second specification, the second quantity.

31. The device according to any one of claims 20 to 24, characterized in that in, The image feature determination module includes: A portrait feature dimension determination submodule, used to determine at least one of a relationship between application indicator data and resources, a topological relationship between resource deployment units, and available resources of a node according to the application indicator data; The portrait feature determination submodule is used to determine the portrait feature based on at least one of the relationship between the application indicator data and resources, the topological relationship between resource deployment units, and the available resources of the node.

32. The device according to claim 29, characterized in that in, The demand resource determination submodule includes: a target resource amount determination unit, configured to determine a target resource amount that matches the application indicator data at the future moment, wherein the target resource amount is determined based on a relationship between the application indicator data and resources; The required resource determination unit is used to determine the required specifications and required quantities of the resource deployment unit according to the target resource quantity.

33. The device according to claim 32, characterized in that in, The target resource quantity determination unit comprises: A resource matching subunit, configured to determine at least one resource quantity data matching the application indicator data at the future moment according to the relationship between the application indicator data and the resource; The target resource quantity determination subunit is used to determine the target resource quantity with the best cost performance ratio from the at least one resource quantity data, wherein the best cost performance ratio is determined based on the minimum value of the ratio between the resource and the application indicator data.

34. The device according to any one of claims 20 to 24, characterized in that in, The resource management module includes: An adjustment submodule is used to adjust at least one of the following according to the linkage strategy to perform resource management: the number of nodes, the specifications of the resource deployment units, and the number of the resource deployment units, wherein the specifications of the resource deployment units, the number of the resource deployment units, and each of the nodes are related to the resources, the resource deployment units are deployed on the nodes, and the cluster includes at least one of the nodes.

35. The device according to claim 34, characterized in that in, The adjustment submodules include: A node resource calling unit is used to call the resources of a target node among candidate nodes to adjust the specifications of the resource deployment unit and / or the number of the resource deployment units, wherein the resources of the target node meet the resource requirements of the specifications of the resource deployment unit and / or the number of the resource deployment units, and the remaining resources of the target node after the resources are called are the smallest among the candidate nodes.

36. The device according to any one of claims 20 to 24, characterized in that in, The resource management module also includes: The linkage policy instruction determination submodule is used to determine the linkage policy instruction according to the linkage policy, wherein the linkage policy instruction is used to instruct the execution of the linkage policy to complete resource management.

37. The device according to any one of claims 20 to 24, characterized in that in, The application indicator data includes at least one of operation indicator data and application characteristic data, wherein the operation indicator data includes at least one of the following: storage resource occupancy data, computing resource occupancy data, event transmission rate and traffic processing rate; the application characteristic data includes at least one of the following: operation sensitivity coefficient, storage sensitivity coefficient and delay sensitivity coefficient.

38. The device according to claim 28, characterized in that in, At least one of the first model, the second model, and the preset indicator supports maintenance.

39. An electronic device comprising: At least one processor, the at least one processor is used to execute instructions so that the electronic device can perform the method according to any one of claims 1-19.

40. A computer-readable storage medium storing computer instructions, wherein: When the computer instructions are executed, the computer is caused to perform the method according to any one of claims 1-19.

Images