Method, system and device for heat migration characteristic resource scheduling decision-making and heat migration

Abstract

The present invention discloses a method, a system and a device for hot migration characteristic resource scheduling decision-making and hot migration. The method includes: selecting a destination physical machine from each alternative destination physical machine according to a first set of hot migration characteristics of a source physical machine, and hot migration characteristics supported by each alternative destination physical machine capable of supporting the first set of hot migration characteristics and available resource information; determining a gap in the hot migration performance between the source physical machine and the destination physical machine; selecting, according to the available resource information of the source physical machine and the destination physical machine, an optimal set of hot migration characteristics capable of filling the gap from the first set of hot migration characteristics as a second set of hot migration characteristics; and determining, according to the second set of hot migration characteristics, resources to be scheduled for each migration characteristic in the second set of hot migration characteristics in the source physical machine and the destination physical machine. The present invention can improve the hot migration performance by using hot migration characteristics during the hot migration process and avoid resource contention at the same time.

Description

Technical Field

[0001] The present invention relates to the field of cloud computing technology, and in particular to a method, system and device for hot migration characteristic resource scheduling decision and hot migration. Background Art

[0002] Hot migration is a data transmission technology. In the field of hot migration technology, for virtual machines under high load, improving the performance of hot migration, for example, the transmission throughput at both ends of the physical machine is higher than the speed of generating dirty pages of the virtual machine memory to complete the convergence of memory delta, is one of the key optimization directions to improve the success rate of hot migration.

[0003] Improving hot migration performance requires some resource-intensive features. These features are usually abstracted from the functions provided to improve hot migration capabilities. There can be multiple features to improve hot migration performance from different dimensions. The technical solutions implemented by these features rely on the resources on the physical machine, that is, during the hot migration process, additional resources such as CPU, memory, bandwidth, etc. need to be consumed from the physical machine. For the sake of convenience, these features are referred to as hot migration features below.

[0004] Due to the resource occupation factors of these hot migration features, using hot migration features is likely to cause resource contention, leading to problems such as virtual machine performance degradation and user business damage. Therefore, on the basis of improving the success rate of hot migration, higher requirements are placed on the resource allocation management of hot migration features. Summary of the invention

[0005] In view of the above problems, the present invention is proposed to provide a method, system and device for hot migration characteristic resource scheduling decision and hot migration that overcomes the above problems or at least partially solves the above problems.

[0006] In a first aspect, an embodiment of the present invention provides a hot migration characteristic resource scheduling decision method, comprising:

[0007] Selecting a destination physical machine from the candidate destination physical machines according to the first hot migration feature set of the source physical machine and the hot migration features and available resource information supported by the candidate destination physical machines that can support the first hot migration feature set;

[0008] Determine the hot migration performance gap between the source and destination physical machines;

[0009] According to the available resource information of the source end physical machine and the destination end physical machine, selecting the best hot migration characteristic set that can make up for the gap from the first hot migration characteristic set as the second hot migration characteristic set;

[0010] Determine the resources that need to be scheduled for each migration characteristic in the second heat migration characteristic set in the source physical machine and the destination physical machine according to the second heat migration characteristic set.

[0011] In one embodiment, according to the first heat migration characteristic set of the source physical machine, and the heat migration characteristics and available resource information supported by each alternative destination physical machine that can support the first heat migration characteristic set, select a destination physical machine from the alternative destination physical machines, including:

[0012] Filter out the alternative destination physical machines that cannot support any heat migration characteristics in the first heat migration characteristic set;

[0013] For the source physical machine and different alternative destination physical machines, according to the minimum required resource amount and the maximum required resource amount corresponding to the heat migration characteristics in the first heat migration characteristic set, and the available resource amounts of different alternative destination physical machines, select a destination physical machine from the alternative destination physical machines according to a preset weight algorithm.

[0014] In one embodiment, determining the gap in the heat migration performance between the source physical machine and the destination physical machine includes:

[0015] Obtain the available resource information of the source physical machine and the destination physical machine;

[0016] Classify the improvable performance defined by each heat migration characteristic in the first heat migration characteristic set, and analyze the improvable performance parameter values of the source physical machine and the destination physical machine in each classification according to the available resource information of the source physical machine and the destination physical machine and the classification, as the gap in the heat migration performance.

[0017] In one embodiment, according to the available resource information of the source physical machine and the destination physical machine, select the best heat migration characteristic set that can make up the gap from the first heat migration characteristic set as the second heat migration characteristic set, including:

[0018] Determine various possible combinations of each heat migration characteristic in the first heat migration characteristic set;

[0019] Take the improvable performance parameter value of each category as a single empty box, and according to a preset bin-packing algorithm, add points when the performance is increased when the gap is not met, and subtract points when the performance is increased after the gap is met, calculate the evaluation score corresponding to each combination, and take the combination of heat migration characteristics with the maximum evaluation score as the second heat migration characteristic set.

[0020] In one embodiment, the first heat migration characteristic set of the source physical machine is determined by the following method:

[0021] Determine the intersection of the live migrations currently supported by the source physical machine and the set of features expected to be enabled for the current live migration input;

[0022] According to the minimum required resource amounts of each first live migration feature in the intersection and the available resources of the source physical machine, determine from the intersection the live migration features that satisfy the condition that the minimum required resource amount is less than the available resources of the source physical machine, as the first set of live migration features.

[0023] In a second aspect, an embodiment of the present invention provides a method for live migration, including:

[0024] Determine the second set of live migration features that need to be enabled for the source physical machine and the destination physical machine during the live migration process and the resources that need to be scheduled;

[0025] Enable the live migration features in the second set of live migration features for the source physical machine and the destination physical machine, and use the resources that need to be scheduled to perform the live migration operation of the source physical machine to the destination physical machine;

[0026] The determination of the second set of live migration features that need to be enabled for the source physical machine and the destination physical machine during the live migration process and the resources that need to be scheduled is determined by the live migration feature resource scheduling decision method in the first aspect above.

[0027] In one embodiment, the step of using the resources that need to be scheduled to perform the live migration operation of the source physical machine to the destination physical machine includes:

[0028] Create corresponding resource reservation instances according to the resources that need to be scheduled for each migration feature in the second set of live migration features in the source physical machine and the destination physical machine;

[0029] Input the identifier of the resource reservation instance and the attribute information of the resource reservation instance to the virtualization layer; the attribute information of the resource reservation instance includes: reserved resource amount, resource reservation timeout time, and identifier of the physical machine where the resource reservation is located;

[0030] Through the virtualization layer, bind the resources corresponding to the resource reservation instances for the live migration operation and use the bound resources during the live migration operation.

[0031] In one embodiment, after the live migration operation is completed, it further includes:

[0032] Release the resource reservation instances according to the timed release policy defined by the resource reservation timeout time;

[0033] And poll to detect whether there are resource reservation instances that have not been released due to timeout;

[0034] If it is detected that there is an existence, release the resource reservation instance.

[0035] In a third aspect, an embodiment of the present invention provides a resource scheduling decision system for hot migration characteristics, including:

[0036] A hot migration characteristics decision scheduling service module, configured to determine a second set of hot migration characteristics to be enabled on the source physical machine and the destination physical machine during the hot migration process and the resources to be scheduled according to the hot migration characteristics information of the physical machines in the physical machine resource pool stored by the hot migration characteristics metadata management service module and the available resource information of the physical machines, in accordance with the hot migration characteristics resource scheduling decision method as described above;

[0037] A hot migration characteristics metadata management service module, configured to collect and store the hot migration characteristics information of the physical machines in the physical machine resource pool;

[0038] A physical machine resource management service module, configured to manage the available resource information of the physical machines in the physical machine resource pool.

[0039] In a fourth aspect, an embodiment of the present invention provides a hot migration service system, including:

[0040] A hot migration characteristics decision scheduling service module, configured to determine a second set of hot migration characteristics to be enabled on the source physical machine and the destination physical machine during the hot migration process and the resources to be scheduled according to the hot migration characteristics resource scheduling decision method as described above;

[0041] A resource reservation instance management service module, configured to create corresponding resource reservation instances according to the resources to be scheduled at both ends of the source physical machine and the destination physical machine;

[0042] A virtualization layer module, configured to bind the resources corresponding to the resource reservation instances to the resource reservation instances and use the bound resources during the hot migration operation;

[0043] A hot migration characteristics metadata management service module, configured to collect and store the hot migration characteristics information of the physical machines in the physical machine resource pool;

[0044] A physical machine resource management service module, configured to manage the available resource information of the physical machines in the physical machine resource pool.

[0045] In a fifth aspect, an embodiment of the present invention provides a server, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the hot migration characteristics resource scheduling decision method as described above, or implements the hot migration method as described above.

[0046] In a sixth aspect, an embodiment of the present invention provides a cloud computing system, including: at least one server as described above, and at least two cloud computing devices supporting live migration;

[0047] Among the at least two cloud computing devices supporting live migration, there are a source physical machine and a destination physical machine that perform live migration.

[0048] In a seventh aspect, an embodiment of the present invention provides a computer-readable storage medium storing a computer program, which when executed by a processor, implements the live migration characteristic resource scheduling decision method as described above, or implements the live migration method as described above.

[0049] In an eighth aspect, an embodiment of the present invention provides a computer program product, which includes a computer program that when executed by a processor, implements the live migration characteristic resource scheduling decision method as described above, or implements the live migration method as described above.

[0050] The beneficial effects of the above technical solutions provided by the embodiments of the present invention at least include:

[0051] The live migration characteristic resource scheduling decision method, live migration method, and related systems and devices provided by the embodiments of the present invention, through unified scheduling of resources, supported live migration characteristics, and available resources on each physical machine, select a destination physical machine, and based on the selected destination physical machine, determine the gap in live migration performance between the two ends, and then select the optimal set of live migration characteristics to be started during the live migration process for both sides, and based on this set, calculate the resources required to enable these live migration characteristics. The embodiments of the present invention achieve reasonable and flexible arrangement and scheduling of the live migration characteristics to be started and the resources to be reserved during the live migration process, thereby avoiding problems such as resource contention during the hot start process, ensuring the improvement of live migration performance in the case of using live migration characteristics, promoting the improvement of the coverage rate of live migration characteristics from the side, and at the same time improving the success rate of live migration operations.

[0052] The live migration method provided by the embodiments of the present invention, on the basis of realizing precise scheduling of live migration characteristics and the resources required for live migration characteristics, actively monitors the abnormality of resource occupancy after the live migration operation is completed, and actively releases resources when the resources are not normally released, avoiding the ineffective occupancy of resources and enhancing the flexibility of resource use.

[0053] Other features and advantages of the present invention will be described in the following specification, and, in part, will be obvious from the specification, or will be understood by implementing the present invention. The objectives and other advantages of the present invention can be achieved and obtained by the structures specifically pointed out in the written specification, claims, and drawings.

[0054] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Description of the Drawings

[0055] The accompanying drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention, but do not constitute a limitation to the present invention. In the drawings:

[0056] Figure 1 is a flowchart of the thermal migration characteristic resource scheduling decision method provided by an embodiment of the present invention;

[0057] Figure 2 is a flowchart of the steps for determining the gap in the thermal migration performance of the source physical machine and the destination physical machine in an embodiment of the present invention;

[0058] Figure 3 is a flowchart of selecting the second thermal migration feature set in an embodiment of the present invention;

[0059] Figure 4 is a flowchart of the thermal migration method in an embodiment of the present invention;

[0060] Figure 5 is a flowchart of the steps for performing the thermal migration operation in an embodiment of the present invention;

[0061] Figure 6 is a schematic diagram of the framework for thermal migration characteristic resource management and scheduling in an embodiment of the present invention;

[0062] Figure 7 is a flowchart of the interaction between modules in the framework for thermal migration characteristic resource management and scheduling in an embodiment of the present invention;

[0063] Figure 8 is a flowchart of the interaction between modules during the abnormal detection and handling process in an embodiment of the present invention;

[0064] Figure 9 is a block diagram of the structure of the thermal migration characteristic resource scheduling decision system in an embodiment of the present invention;

[0065] Figure 10 is a block diagram of the structure of the thermal migration service system in an embodiment of the present invention. Detailed Embodiments

[0066] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art.

[0067] In the prior art, starting the live migration feature during the live migration process may cause resource contention, leading to problems such as a decline in virtual machine performance and damage to user services. In the prior art, it is also possible to simply use the available resources remaining in the physical machine that are not occupied by virtual machines during live migration or directly shut down the use of the remaining resources on the physical machine, apply for a fixed resource pool, and hand over the fixed resource pool to the live migration function for use. However, this resource scheduling method is rigid and not flexible enough, and the resources cannot be reused, resulting in low resource utilization.

[0068] The inventors of the present invention have found that if the capabilities provided by the live migration feature, that is, the information on relevant resource consumption, are managed, and the resources required by the virtual machine and the live migration feature are uniformly scheduled during each live migration process, and the reuse of the resources required by the live migration feature is realized, it is possible to avoid resource contention during the live migration process, improve the utilization rate of the live migration feature, and thus increase the success rate of live migration.

[0069] Based on this, the embodiments of the present invention provide a method for making a resource scheduling decision for a live migration feature and a method for live migration. Below, with reference to the accompanying drawings, the above-mentioned method for making a resource scheduling decision for a live migration feature and the method for live migration will be described in detail.

[0070] The method for making a resource scheduling decision for a live migration feature provided by the embodiments of the present invention, as shown in Figure 1 includes:

[0071] S11. Select a destination physical machine from each alternative destination physical machine according to the first live migration feature set of the source physical machine, and the live migration features supported by each alternative destination physical machine that can support the first live migration feature set and the available resource information.

[0072] S12. Determine the gap in the live migration performance between the source physical machine and the destination physical machine.

[0073] S13. Select the best live migration feature set that can make up for the gap from the first live migration feature set as the second live migration feature set according to the available resource information of the source physical machine and the destination physical machine.

[0074] S14. Determine the resources that need to be scheduled for each migration feature in the second live migration feature set in the source physical machine and the destination physical machine according to the second live migration feature set.

[0075] The above-mentioned resource scheduling decision method for heat migration characteristics provided by the embodiments of the present invention uniformly schedules the resources, supported heat migration characteristics, and available resources on each physical machine, selects the destination physical machine, determines the gap in the heat migration performance between the two ends based on the selected destination physical machine, and then selects the optimal set of heat migration characteristics that need to be started during the heat migration process for both parties. Based on this set, the required resources are calculated. The embodiments of the present invention achieve reasonable and flexible arrangement and scheduling of the heat migration characteristics that need to be started during the heat migration process and the resources that need to be reserved, thereby avoiding problems such as resource contention during the hot start process, ensuring the improvement of heat migration performance when using heat migration characteristics, promoting the improvement of the coverage rate of heat migration characteristics indirectly, and also improving the success rate of heat migration operations.

[0076] In the embodiments of the present invention, the heat migration characteristics can be abstracted in advance through the capabilities provided by the characteristics. For example:

[0077] 1. For a certain characteristic, if its technical solution can reduce the total amount of memory dirty page (the modified page is called a dirty page) data transmitted over the network and provide compression capabilities or resource data, then a data compression characteristic is generated according to this capability abstraction;

[0078] 2. For a certain characteristic, if its technical solution can improve the heat migration bandwidth input / output (I / O) performance and provide bandwidth capabilities / resource data, then a transmission I / O improvement characteristic is generated according to this capability abstraction.

[0079] Another example is characteristics related to improving network forwarding capabilities, etc. The embodiments of the present invention are not limited to any function that can improve heat migration capabilities, that is, the definition and type of the heat migration characteristics themselves are not limited.

[0080] Based on the definition of each heat migration characteristic, in the embodiments of the present invention, the minimum and maximum required resources for each heat migration characteristic are also defined; the minimum required resources represent the lowest resource quantity required when the heat migration characteristic is enabled, and the maximum required resources represent the highest point of available resources or the resource quantity at the μ position of the normal distribution of the achievable performance (i.e., the highest point of the benefit achieved after using this characteristic). In the embodiments of the present invention, resources include but are not limited to bandwidth, CPU, and other resources.

[0081] The above-mentioned first set of heat migration characteristics of the source physical machine can be obtained through the following method:

[0082] Determine the intersection of the heat migration currently supported by the source physical machine and the set of characteristics expected to be enabled for the current heat migration input;

[0083] According to the minimum required resources of each first heat migration characteristic in the intersection set, and the available resources of the source physical machine, determine from the intersection set the heat migration characteristics that satisfy the condition that the minimum required resources are less than the available resources of the source physical machine, and use them as the first heat migration characteristic set.

[0084] The intersection of the sets of characteristics expected to be enabled for this heat migration can be, for example, the set of heat migration characteristics that the heat migration operator specifies to be enabled, and these sets are input by the heat migration operator.

[0085] Since multiple heat migration characteristics share reserved resources, at least the minimum required resources of the heat migration characteristics should be less than the available resources of the source physical machine. Otherwise, these heat migration characteristics cannot be used.

[0086] Further, in the above S11, according to the first heat migration characteristic set of the source physical machine, and the heat migration characteristics and available resource information supported by each alternative destination physical machine that can support the first heat migration characteristic set, select a destination physical machine from the alternative destination physical machines. Specifically, it can be implemented in the following way:

[0087] Filter out the alternative destination physical machines that cannot support any heat migration characteristics in the first heat migration characteristic set;

[0088] For the source physical machine and different alternative destination physical machines, according to the minimum and maximum required resources corresponding to the heat migration characteristics in the first heat migration characteristic set, and the available resources of different alternative destination physical machines, select a destination physical machine from the alternative destination physical machines according to a preset weight algorithm.

[0089] For each different alternative destination physical machine, the heat migration characteristics it can support may be different, so the maximum and minimum required resources for these heat migration characteristics may also be different. If the source physical machine and an alternative destination physical machine form a combination, then for these combinations, the available resources of the destination physical machine are also different. Considering from the available resources and the supported characteristics, set different weight values for different characteristics, and also set corresponding weight values for the maximum and minimum required resources of different characteristics. Considering comprehensively and using a weight algorithm, comprehensively consider aspects such as the function of the characteristics, the maximum and minimum resources required by the supported characteristics, and the available resources of the destination physical machine, and select from the alternative destination physical machines a destination physical machine that can achieve as many and as important heat migration characteristics as possible with as little resource as possible.

[0090] The specific weight algorithm can be implemented using various existing algorithms, and is not limited here.

[0091] In one embodiment, after selecting the destination physical machine, in step S12 above, determine the gap in the live migration performance between the source physical machine and the destination physical machine. Refer to Figure 2 as shown, it can be implemented through the following process:

[0092] S21. Obtain the available resource information of the source physical machine and the destination physical machine;

[0093] S22. Classify the improvable performance defined by each live migration characteristic in the first set of live migration characteristics;

[0094] S23. According to the available resource information of the source physical machine and the destination physical machine, and the classification, analyze the improvable performance parameter values of the source physical machine and the destination physical machine in each classification, and use them as the gap in the live migration performance.

[0095] For different live migration performances, different technical solutions are used to improve the live migration performance. Therefore, according to the definitions of each live migration characteristic, their improvable performances can be classified, such as improving data compression ability, improving transmission ability, etc. Then, according to the available resource information of the source physical machine and the destination physical machine, determine the improvable performance parameter values in each classification, such as bandwidth value, I / O rate, memory compression rate per unit time, etc.

[0096] In the embodiment of the present invention, the gap can be a gap comprehensively formed by the improvable performance parameter values in multiple classifications, and in specific implementation, a performance gap model can be used to represent it.

[0097] An example of the performance gap model is as follows:

[0098] Examples of the improvable performance parameter values defined by this gap model under several different classifications:

[0099] 1) The maximum available bandwidth of the current live migration dual-end physical machines (source and destination physical machines) = (total physical machine network bandwidth - used bandwidth on the physical machine (i.e., the allocated bandwidth));

[0100] 2) The maximum available compression ability of the current live migration dual-end physical machines (source and destination physical machines) (total physical machine HT - sold HT of the physical machine) Compression rate provided by a single HT; the above HT is the number of hyper-threads (HT, Hyper-Threading).

[0101] 3) Others.

[0102] Through the multi-dimensional improvable performance parameter values, a performance gap model can be defined.

[0103] In one embodiment, in step S13 above, refer toFigure 3 As shown in the figure, the second set of heat migration features can be selected in the following manner:

[0104] S31. Determine all possible combinations of the heat migration characteristics in the first set of heat migration characteristics;

[0105] S32. Take the performance parameter value that can be improved for each category as a single empty bin. According to the preset bin-packing algorithm, if the notch is not met and the performance is increased, points are added; if the notch is met and the performance is increased, points are deducted. Calculate the evaluation score corresponding to each combination, and take the combination of heat migration characteristics with the highest evaluation score as the second set of heat migration characteristics.

[0106] In the bin-packing algorithm, take the performance parameter value that can be improved for each category as a single empty bin. Each empty bin has a corresponding ability to fill a certain notch. Use the empty bins to "load" the notch in sequence, calculate the evaluation scores under various combinations, and take the combination of heat migration characteristics corresponding to the maximum score as the final second set of heat migration characteristics.

[0107] For example, for all possible combinations of heat migration characteristics, in the bin-packing algorithm, the heat migration characteristics have a sequence. Therefore, in some cases, even if the combinations contain the same several heat migration characteristics, they are regarded as different combinations because of the different sequences. For example, ABC and ACB are two different combinations.

[0108] Process of the bin-packing algorithm: If the ability of A is not enough to fill the notch, then when adding A, the evaluation score uses a positive value, that is, the total evaluation score increases. Continuing to put in C, the total evaluation score continues to increase. If the notch can be filled at this time, then continuing to put in B, the total evaluation score will start to decrease, that is, B corresponds to a negative score. From this, it can be known that the evaluation score of the combination ACB is lower than that of the combination AC.

[0109] Referring to the above method, the evaluation scores of various combinations can be calculated respectively. From this, select the combination of heat migration characteristics corresponding to the maximum score as the second set of heat migration characteristics, that is, the best combination closest to the performance notch.

[0110] According to the second set of heat migration characteristics, the resource information that the source physical machine and the destination physical machine need to reserve for the heat migration characteristics in the second set of heat migration characteristics can be calculated.

[0111] The above-mentioned resource scheduling decision method for hot migration characteristics provided by the embodiments of the present invention can perform scheduling according to the available resources of the source physical machine and the destination physical machine and various hot migration characteristics supported, and can uniformly classify and calculate the optimal combination of hot migration characteristics and the corresponding reserved resources during the hot migration process. Compared with the prior art in which the use of the remaining resources on the physical machine needs to be shut down when using hot migration characteristics to prevent resource contention, the flexibility of the use of available resources on the physical machine can be improved, making the resource utilization rate higher, thereby indirectly improving the coverage rate of hot migration characteristics and the success rate of hot migration.

[0112] Based on the same inventive concept, the embodiments of the present invention also provide a hot migration method, as shown in Figure 4 and including:

[0113] S41. Determine the second set of hot migration characteristics that need to be enabled on the source physical machine and the destination physical machine during the hot migration process, and the resources that need to be scheduled;

[0114] S42. Enable the hot migration characteristics in the second set of hot migration characteristics on the source physical machine and the destination physical machine, and use the resources that need to be scheduled to perform the hot migration operation from the source physical machine to the destination physical machine;

[0115] The determination of the second set of hot migration characteristics that need to be enabled on the source physical machine and the destination physical machine during the hot migration process, and the resources that need to be scheduled, is determined by the resource scheduling decision method for hot migration characteristics provided in the foregoing embodiments.

[0116] Further, in the above step 42, using the resources that need to be scheduled to perform the hot migration operation from the source physical machine to the destination physical machine, as shown in Figure 5 and can be implemented through the following process:

[0117] S51. Create corresponding resource reservation instances according to the resources that need to be scheduled for each migration characteristic in the second set of hot migration characteristics on the source physical machine and the destination physical machine;

[0118] S52. Input the identifier of the resource reservation instance and the attribute information of the resource reservation instance to the virtualization layer; the attribute information of the resource reservation instance includes: the reserved resource amount, the resource reservation timeout time, and the identifier of the physical machine where the resource reservation is located;

[0119] S53. Through the virtualization layer, bind the resources corresponding to the resource reservation instance for the hot migration operation, and use the bound resources during the hot migration operation.

[0120] Before live migration, it is necessary to create a resource reservation instance for live migration. This resource reservation instance corresponds to the actual physical resources on the physical machine and stipulates information such as the resource occupancy and the resource reservation timeout. Through the virtualization layer, during the live migration process, these reserved resources can be bound to the live migration features in the enabled second live migration feature set for use, so as to prevent resource contention and damage to user services during the live migration process.

[0121] In order to meet the purpose of flexible resource use, after the live migration operation is completed, the resource reservation instance can also be released according to the timed release policy defined by the resource reservation timeout; and it is polled to detect whether there is a resource reservation instance that has not been released after timeout; if it is detected that there is one, the resource reservation instance is released.

[0122] In some cases, it may occur that after the live migration ends and the timeout has occurred, the resource instance is not released. Then, it is necessary to check the live migration status and confirm that the resources are released after the live migration ends to avoid the situation of invalid resource occupation.

[0123] To better illustrate the above-mentioned live migration feature resource scheduling decision method and the live migration method provided by the embodiments of the present invention, a specific example is used for illustration.

[0124] This example provides a framework for live migration feature resource management and scheduling as Figure 6 shown. In this framework, in the live migration feature resource management and scheduling framework, it includes but is not limited to the following several modules:

[0125] Live migration feature metadata management service (MetaService), physical machine resource management service (ResourceService), resource reservation instance management service (ReserveInstanceService), live migration feature decision scheduling service (ScheduleService), live migration progress management service (MigrationService), etc.

[0126] By building a complete set of processes from live migration feature metadata collection, physical machine resource situation analysis, live migration feature scheduling and corresponding resource application, and exception status handling, the live migration memory delta convergence performance is improved, thereby increasing the success rate of live migration.

[0127] The live migration feature metadata management service (MetaService) collects the information of the live migration features supported on each physical machine (including the physical machine where the virtual machine is currently hosted and several optional live migration destination physical machines) in the physical machine resource layer through methods such as virtualization software version collection and virtualization API interfaces, classifies them according to the model, and then performs persistent storage.

[0128] The hot migration feature metadata includes: the capability information provided by the hot migration feature, and the minimum and maximum required resource information of the hot migration feature.

[0129] The hot migration features can be classified according to the models with reference to the capabilities provided by the various hot migration features described above, which will not be elaborated here.

[0130] As the detection, collection, and global management entry of physical machine resource information, the Physical Machine Resource Management Service (ResourceService) can perform resource query, application, and return services for a single physical machine resource in a distributed system through means such as distributed locks, database transactions, resource allocation algorithms, and automatic repair in case of exceptions. It interacts with the Resource Reservation Instance Management Service (ResourceInstanceService) to realize the scheduling of reserved resources for the hot migration feature within the framework.

[0131] Based on the metadata of the hot migration feature collected by the Hot Migration Feature Decision Scheduling Service (ScheduleService) and combined with the available resource information of the physical machine, etc., it makes decision scheduling for the actually enabled hot migration features of a single hot migration.

[0132] As the middle layer between the hot migration service and the resource management service, the Resource Reservation Instance Management Service (ReserveInstanceService) provides a reserved instance model and corresponding life cycle management services: for example, creating a reserved instance before hot migration, actively releasing the reserved instance when the hot migration ends, configuring a timed release policy, etc. At the same time, combined with the record of the execution of the entire hot migration cycle by the Hot Migration Progress Management Service (MigrationService), it can achieve active monitoring and automatic repair in case of unexpected exceptions during hot migration to ensure the availability of the remaining resources of the physical machine.

[0133] After completing the decision-making on the actually enabled hot migration features and resources, the Hot Migration Feature Decision Scheduling Service (ScheduleService) hands over the application for the required resources to the Resource Reservation Instance Management Service (ResourceInstanceService).

[0134] The Resource Reservation Instance Management Service (ResourceInstanceService) will first abstract the resources to be applied for into resource reservation instances in its internal logic, so that the hot migration resources and virtual machine resources can be uniformly orchestrated and managed by the Physical Machine Resource Management Service (ResourceService).

[0135] After the scheduling decision is made by the Heat Migration Feature Decision Scheduling Service (ScheduleService), the Heat Migration Progress Management Service (MigrationService) summarizes information such as the set of enabled heat migration features, information on resource reservation instances, resource allocation information, etc., and records it in the persistent additional information corresponding to this heat migration. At the same time, it is responsible for interacting with the virtualization layer ( Figure 6 not shown in the figure, located Figure 6 between the heat migration feature resource management and scheduling framework and the underlying physical machine resource layer in

[0136] Refer to Figure 7 As shown, the interaction process of each module in the above framework for scheduling the resources required for the heat migration feature and during the heat migration process is as follows:

[0137] 1. The Heat Migration Feature Metadata Management Service requests the physical machine heat migration feature metadata;

[0138] 2. The virtualization layer returns the metadata;

[0139] 3. The Heat Migration Progress Management Service requests the destination physical machine for scheduling and the decision heat migration features, and inputs the specified set of heat migration feature information (if it supports the external user to specify an optional set of heat migration features, the specified set of heat migration features can be input);

[0140] 4. The Heat Migration Feature Decision Scheduling Service requests the set of features currently supported by the source host;

[0141] 5. The Heat Migration Feature Metadata Management Service returns the set of features currently supported;

[0142] 6. The Heat Migration Feature Decision Scheduling Service requests the current remaining resources of the source host;

[0143] 7. The Physical Machine Resource Management Service returns the current remaining resources;

[0144] 8. The Heat Migration Feature Decision Scheduling Service calculates the set of features that can be enabled to meet the minimum resource amount;

[0145] 9. The Heat Migration Feature Decision Scheduling Service obtains the set result;

[0146] 10. The Heat Migration Feature Decision Scheduling Service filters out the alternative destination ends that do not support any features in the set;

[0147] 11. The Heat Migration Feature Decision Scheduling Service determines the set of destination physical machines that support at least one optional feature;

[0148] 12. The heat migration characteristic decision scheduling service uses a weight algorithm to select the optimal destination that can best meet the optional characteristics;

[0149] 13. The heat migration characteristic decision scheduling service finally selects the destination;

[0150] 14. The heat migration characteristic decision scheduling service makes a decision on the heat migration characteristic;

[0151] 15. The heat migration characteristic decision scheduling service finally selects the set of heat migration characteristics and the resource reservation information required for both ends;

[0152] 16. The heat migration characteristic decision scheduling service requests to create the scheduled resource application as a resource reservation instance;

[0153] 17. The resource reservation instance management service converts the resource into a resource reservation instance;

[0154] 18. The resource reservation instance management service obtains the resource reservation instance;

[0155] 19. The resource reservation instance management service occupies the resource through the resource reservation instance interaction;

[0156] 20. The resource reservation instance management service obtains the detailed reserved resource metadata;

[0157] 21. The resource reservation instance management service returns the metadata of the resource reservation instance and the detailed reserved resource metadata to the heat migration characteristic decision scheduling service;

[0158] 22. The heat migration characteristic decision scheduling service decides on the set of heat migration characteristics to be enabled, the ID of the reservation instance, and the detailed reserved resource metadata, and sends them to the heat migration progress management service;

[0159] 23. The heat migration progress management service persists the obtained set of heat migration characteristics, the ID of the reservation instance, the detailed reserved resource metadata, etc.;

[0160] 24. The heat migration progress management service notifies the virtualization layer to perform heat migration and inputs the detailed reserved resource metadata;

[0161] 25. The virtualization layer binds the resource for heat migration, prevents contention, and performs heat migration;

[0162] 26. Heat migration is completed;

[0163] 27. The heat migration progress management service applies to release the resource reservation instance;

[0164] 28. The resource reservation instance management service completes the release of the resource reservation instance.

[0165] After the resource reservation instance is released, refer to Figure 8 As shown, it also includes the following step anomaly detection and processing process:

[0166] 1. The virtualization layer automatically releases the bound resources;

[0167] 2. The resource reservation instance management service looks up the records of resource reservation instances that have expired but have not been actively released;

[0168] 3. Find the corresponding records;

[0169] 4. Confirm the migration status;

[0170] 5. Confirm that the migration has ended;

[0171] 6. The resource reservation instance management service notifies the physical machine resource management service to release the occupied resources;

[0172] 7. The resource release is completed.

[0173] In step 6 above, the physical resource management service can interact with the virtualization layer to release the physical resources occupied by the resource reservation instance.

[0174] Based on the same inventive concept, the embodiments of the present invention also provide a hot migration characteristic resource scheduling decision system, a hot migration service system, a server, and a cloud computing system. Since the principles of the problems solved by these systems and servers are similar to those of the foregoing hot migration characteristic resource scheduling decision method and the method of hot migration, the implementation of these systems, servers, and cloud computing systems can refer to the implementation of the foregoing methods, and the repeated parts will not be elaborated.

[0175] The embodiments of the present invention provide a hot migration characteristic resource scheduling decision system. As shown in Figure 9, it includes:

[0176] The hot migration characteristic decision scheduling service module 91 is used to determine the second set of hot migration characteristics to be enabled and the resources to be scheduled for the source physical machine and the destination physical machine during the hot migration process according to the hot migration characteristic information of the physical machines in the physical machine resource pool stored by the hot migration characteristic metadata management service module and the available resource information of the physical machines, in accordance with the foregoing hot migration characteristic resource scheduling decision method;

[0177] The hot migration characteristic metadata management service module 92 is used to collect and store the hot migration characteristic information of the physical machines in the physical machine resource pool;

[0178] The physical machine resource management service module 93 is used to manage the available resource information of the physical machines in the physical machine resource pool.

[0179] The embodiments of the present invention provide a hot migration service system. As shown in Figure 10, it includes:

[0180] The thermal migration characteristic decision scheduling service module 101 is used to determine the second set of thermal migration characteristics to be enabled on the source physical machine and the destination physical machine and the resources to be scheduled during the thermal migration process according to the thermal migration characteristic resource scheduling decision method as described above.

[0181] The resource reservation instance management service module 102 is used to create corresponding resource reservation instances according to the resources to be scheduled at both the source physical machine and the destination physical machine.

[0182] The virtualization layer module 103 is used to bind the resources corresponding to the resource reservation instances to the resource reservation instances and use the bound resources during the thermal migration operation.

[0183] The thermal migration characteristic metadata management service module 104 is used to collect and store the thermal migration characteristic information of the physical machines in the physical machine resource pool.

[0184] The physical machine resource management service module 105 is used to manage the available resource information of the physical machines in the physical machine resource pool.

[0185] An embodiment of the present invention provides a server, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the thermal migration characteristic resource scheduling decision method as described above, or implements the thermal migration method as described above.

[0186] An embodiment of the present invention provides a cloud computing system, including: at least one server as described above, and at least two cloud computing devices supporting thermal migration, wherein these two cloud computing devices supporting thermal migration include a source physical machine and a destination physical machine for performing thermal migration.

[0187] An embodiment of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the thermal migration characteristic resource scheduling decision method as described above, or implements the thermal migration method as described above.

[0188] An embodiment of the present invention provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the thermal migration characteristic resource scheduling decision method as described above, or implements the thermal migration method as described above.

[0189] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) that contain computer-usable program code.

[0190] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0191] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing devices to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that realizes the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0192] These computer program instructions can also be loaded onto a computer or other programmable data processing devices, such that a series of operation steps are executed on the computer or other programmable devices to generate a computer-implemented process, so that the instructions executed on the computer or other programmable devices provide steps for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0193] Obviously, those skilled in the art can make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention also intends to include these modifications and variations.

Claims

1. A resource scheduling decision method for thermal migration characteristics, characterized in that Including: Select a destination physical machine from the alternative destination physical machines according to the first live migration characteristic set of the source physical machine, and the live migration characteristics and available resource information supported by each alternative destination physical machine capable of supporting the first live migration characteristic set; Determine the gap in the live migration performance between the source physical machine and the destination physical machine; The gap in the live migration performance is determined according to the available resource information of the source physical machine and the destination physical machine and the classification of the performance that can be improved defined by each live migration characteristic in the first live migration characteristic set; the gap in the live migration performance is the performance parameter value that can be improved by the source physical machine and the destination physical machine in each classification; Select the best live migration characteristic set that can bridge the gap from the first live migration characteristic set as the second live migration characteristic set according to the available resource information of the source physical machine and the destination physical machine; Determine the resources that need to be scheduled for each migration characteristic in the second live migration characteristic set in the source physical machine and the destination physical machine according to the second live migration characteristic set; 2. The method according to claim 1, wherein Select a destination physical machine from the alternative destination physical machines according to the first live migration characteristic set of the source physical machine, and the live migration characteristics and available resource information supported by each alternative destination physical machine capable of supporting the first live migration characteristic set, including: Filter out the alternative destination physical machines that cannot support any live migration characteristic in the first live migration characteristic set; For the source physical machine and different alternative destination physical machines, according to the minimum required resource amount and the maximum required resource amount corresponding to the live migration characteristics in the first live migration characteristic set, and the available resource amounts of different alternative destination physical machines, select the destination physical machine from the alternative destination physical machines according to a preset weight algorithm; 3. The method according to claim 1, wherein Determine the gap in the live migration performance between the source physical machine and the destination physical machine, including: Obtain the available resource information of the source physical machine and the destination physical machine; Classify the performance that can be improved defined by each live migration characteristic in the first live migration characteristic set, and analyze the performance parameter values that can be improved by the source physical machine and the destination physical machine in each classification according to the available resource information of the source physical machine and the destination physical machine and the classification, as the gap in the live migration performance; 4. The method according to claim 3, wherein Select the best live migration characteristic set that can bridge the gap from the first live migration characteristic set as the second live migration characteristic set according to the available resource information of the source physical machine and the destination physical machine, including: Determine various possible combinations of each live migration characteristic in the first live migration characteristic set; Take the performance parameter value that can be improved in each category as a single empty bin, and according to a preset bin-packing algorithm, add points for increasing performance when the gap is not met and subtract points for increasing performance after the gap is met, calculate the evaluation score corresponding to each combination, and take the combination of live migration characteristics with the maximum evaluation score as the second live migration characteristic set; 5. The method according to any one of claims 1 to 4, characterized in that, The first live migration characteristic set of the source physical machine is determined by the following method: Determine the intersection of the live migrations currently supported by the source physical machine and the set of features expected to be enabled for the current live migration input; According to the minimum required resource amounts of each first live migration feature in the intersection and the available resources of the source physical machine, determine from the intersection the live migration features that satisfy the condition that the minimum required resource amount is less than the available resources of the source physical machine, as the first live migration feature set.

6. A method for thermal migration, characterized in that, Including: Determine the second live migration feature set to be enabled and the resources to be scheduled for the source physical machine and the destination physical machine during the live migration process; Enable the live migration features in the second live migration feature set for the source physical machine and the destination physical machine, and use the resources to be scheduled to perform the live migration operation of the source physical machine to the destination physical machine; The determination of the second live migration feature set to be enabled and the resources to be scheduled for the source physical machine and the destination physical machine during the live migration process is determined by the live migration feature resource scheduling decision method according to any one of claims 1-5.

7. The method according to claim 6, wherein The use of the resources to be scheduled to perform the live migration operation of the source physical machine to the destination physical machine includes: According to the resources to be scheduled for each migration feature in the second live migration feature set in the source physical machine and the destination physical machine, create corresponding resource reservation instances; Input the identifier of the resource reservation instance and the attribute information of the resource reservation instance to the virtualization layer; the attribute information of the resource reservation instance includes: reserved resource amount, resource reservation timeout, and physical machine identifier where the resource reservation is located; Through the virtualization layer, bind the resources corresponding to the resource reservation instance for the live migration operation and use the bound resources during the live migration operation.

8. The method according to claim 7, wherein After the live migration operation is completed, it further includes: Release the resource reservation instance according to the timed release policy defined by the resource reservation timeout; And poll to detect whether there are resource reservation instances that have not been released due to timeout; If it is detected that there are, release the resource reservation instance.

9. A thermal migration characteristic resource scheduling decision-making system, characterized in that Including: The live migration feature decision scheduling service module is used to determine the second live migration feature set to be enabled and the resources to be scheduled for the source physical machine and the destination physical machine during the live migration process according to the live migration feature information of the physical machines in the physical machine resource pool stored by the live migration feature metadata management service module and the available resource information of the physical machines, and according to the live migration feature resource scheduling decision method according to any one of claims 1-5; The live migration feature metadata management service module is used to collect and store the live migration feature information of the physical machines in the physical machine resource pool; The physical machine resource management service module is used to manage the available resource information of the physical machines in the physical machine resource pool.

10. A thermal migration service system, characterized in that, Including: The live migration feature decision scheduling service module is used to determine the second live migration feature set to be enabled and the resources to be scheduled for the source physical machine and the destination physical machine during the live migration process according to the live migration feature resource scheduling decision method according to any one of claims 1-5; A resource reservation instance management service module, which is used to create corresponding resource reservation instances according to the resources to be scheduled at both the source physical machine and the destination physical machine; A virtualization layer module, which is used to bind the resources corresponding to the resource reservation instances to the resource reservation instances and use the bound resources during the live migration operation; A live migration feature metadata management service module, which is used to collect and store the live migration feature information of the physical machines in the physical machine resource pool; A physical machine resource management service module, which is used to manage the available resource information of the physical machines in the physical machine resource pool.

11. A server, characterized in that, It includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the live migration feature resource scheduling decision method according to any one of claims 1-5, or implements the live migration method according to any one of claims 6-8.

12. A cloud computing system, characterized in that, It includes: At least one server as described in claim 11, and at least two cloud computing devices supporting live migration; The at least two cloud computing devices supporting live migration include a source physical machine and a destination physical machine that perform live migration.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the live migration feature resource scheduling decision method according to any one of claims 1-5, or implements the live migration method according to any one of claims 6-8.

14. A computer program product, characterized in that, The computer program product includes a computer program, and when the computer program is executed by a processor, it implements the live migration feature resource scheduling decision method according to any one of claims 1-5, or implements the live migration method according to any one of claims 6-8.

Images