Resource sharing calling method and device based on cloud platform and storage medium

Abstract

The invention relates to the technical field of cloud data processing, in particular to a resource sharing and calling method and device based on a cloud platform and a storage medium, and the method comprises the steps: collecting cloud server resource data, uploading data of each process and node information in real time, classifying the cloud server processes, and storing the classified cloud server processes in a cloud server database; initial computing power is distributed to each process in combination with uploaded data of each process, a real-time shared resource state is determined according to node information of each edge node and cloud server resource data, and the distributed computing power is set according to an analysis result; establishing a relation table between the behavior mode characteristics of each edge node and the load time of the cloud server, and identifying the behavior cycle of each edge node and the load cycle of the cloud server; the resource load state of the next monitoring period is predicted, and then the distributed computing power of each process in the current monitoring period is updated. According to the invention, the calling efficiency of the shared resources is effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cloud data, and particularly relates to a resource sharing calling method and device based on a cloud platform and a storage medium. BACKGROUND

[0002] At present, with the rapid development of cloud computing technology, more and more enterprises and organizations tend to migrate data processing tasks to the cloud. However, in actual application, the cloud platform faces problems such as uneven resource allocation, response delay, and waste of computing power. Especially in the face of sudden high data processing demand, the traditional resource sharing and calling method is difficult to efficiently utilize the resources between the edge nodes and the cloud servers, resulting in low resource utilization and low service quality. In addition, due to the diversity of data types uploaded by each edge node and the complexity of behavior patterns, it is particularly important to understand the behavior cycle of these nodes and the load cycle of the cloud server. However, the existing technology has deficiencies in identifying these periodic characteristics, and cannot accurately predict the resource load state in the next monitoring cycle, thereby affecting the efficiency of resource sharing and calling.

[0003] In order to solve the above problems, some existing technologies attempt to improve resource use efficiency by optimizing resource scheduling algorithms. SUMMARY

[0004] The present application aims to provide a resource sharing calling method and device based on a cloud platform and a storage medium to solve at least one of the problems existing in the prior art.

[0005] To achieve the above-mentioned purpose, according to one aspect of the present application, a resource sharing calling method based on a cloud platform is provided, comprising:

[0006] The initial computing power is allocated to each process in combination with the uploaded data of each process, the real-time shared resource state is determined based on the node information of each edge node and the cloud server resource data, and the allocated computing power is set;

[0007] The behavior pattern characteristics of each edge node and the load time relationship table of the cloud server are established according to the historical data of each edge node and the task log of the cloud server, and the behavior cycle of each edge node and the load cycle of the cloud server are identified;

[0008] The resource load state of the next monitoring cycle is predicted based on the behavior cycle of each edge node and the load cycle of the cloud server, and the allocated computing power of each process in the current monitoring cycle is updated.

[0009] Optionally, the allocation process of the initial computing power is as follows:

[0010] If the jth process is a cloud process, the minimum computing power signal of the jth process in the cloud server is used as the initial computing power of the jth process;

[0011] If the jth process is an externally executable process, the uploaded data of each process is compared in size with the instruction data corresponding to the process, and when the size of the uploaded data of the jth process is greater than or equal to the size of the instruction data of the jth process, the size of the instruction data of the jth process is transmitted to the edge node of the jth process, and then the minimum computing power signal of the jth process in the edge node is used as the initial computing power of the jth process, and the edge node corresponding to the jth process is used as the running device of the jth process; when the size of the uploaded data of the jth process is less than the size of the instruction data of the jth process, the minimum computing power signal of the jth process in the cloud server is used as the initial computing power of the jth process, and the cloud server is used as the running device of the jth process.

[0012] Optionally, the historical data of each edge node and the task log of the cloud server are obtained, and the computing power redundancy of each process is analyzed according to the obtained results;

[0013] The process of calculating the computing power redundancy of each process is as follows:

[0014] The process instruction is matched with the historical process instruction of the edge node where the process is located and the historical process instruction of the cloud server, and the average value of the historical process instruction that matches successfully is used as the computing power redundancy of the process.

[0015] Optionally, the setting process of the allocation computing power of each process is as follows:

[0016] The resource load index of the cloud server is determined by the resource data of the cloud server, and is set as a; the real-time shared resource state is determined by judging the resource load index through the set load judgment threshold: the real-time shared resource state when the resource load index is less than the load judgment threshold is set as a reasonable sharing state, and the real-time shared resource state when the resource load index is greater than or equal to the load judgment threshold is set as an excessive sharing state;

[0017] If the real-time resource sharing state is an excessive sharing state and a1xv1+a2xv2 is less than 40%, the running device of the jth process is updated to the edge node corresponding to the jth process, and the sum of the computing power redundancy and the initial computing power of the jth process is used as the allocation computing power of the jth process; if the real-time resource sharing state is an excessive sharing state and a1xv1+a2xv2 is greater than or equal to 40%, the initial computing power of the jth process is used as the allocation computing power of the jth process; if the real-time resource sharing state is a reasonable sharing state, the initial computing power of the jth process is used as the allocation computing power of the jth process;

[0018] wherein a1 is the memory occupation weight of the edge node corresponding to the jth process, v1 is the memory occupation ratio of the edge node corresponding to the jth process, a2 is the processor utilization of the edge node corresponding to the jth process, and v2 is the processor utilization of the edge node corresponding to the jth process.

[0019] Optionally, the behavior mode features of the edge nodes are determined according to historical data of the edge nodes, and the behavior periods of the edge nodes are identified according to the behavior mode features of the edge nodes; the behavior period is the period of the behavior mode of the data uploaded by the edge nodes.

[0020] The load schedule of the cloud server is established according to the task log of the cloud server, and the load period of the cloud server is identified according to the load schedule.

[0021] Optionally, the access time stamps of the edge nodes are divided into time interval segments at a preset interval time, the behavior indexes of the edge nodes in each time interval segment are determined according to the number of times of accessing the cloud server and the amount of data processed by the edge nodes when accessing the cloud server, and are recorded as xw(i, k), and xw(i, k) is set as the product of the number of times of accessing the cloud server and the amount of data processed by the kth edge node when accessing the cloud server in the ith time interval segment, wherein i and k are both numerical indices, representing the time interval number and the number of edge nodes, respectively.

[0022] Optionally, the coefficient of variation CV(k) and the maximum behavior index max(k) of the behavior index of the kth edge node in the interval period are calculated, and a preset standard is set to determine the behavior mode of the edge nodes: when CV(k) is greater than 40% and max(k) is less than MAX(k), the ith time interval segment and the maximum behavior index max(k) are taken as the behavior mode features of the kth edge node; when CV(k) is less than or equal to 40%, only max(k) is taken as the behavior mode features of the kth edge node, wherein MAX(k) is the kth peak access constant.

[0023] The behavior mode features of the edge nodes are input into an autocorrelation function to obtain the behavior periods of the edge nodes.

[0024] The cloud server load index fz(i) is set in the manner of setting the behavior indexes of the edge nodes in each time interval segment, and the cloud server load indexes of each time interval segment are taken as the load time relationship table.

[0025] The load time relationship table of the cloud server is input into an autocorrelation function to identify the load period of the cloud server.

[0026] Optionally, the behavior pattern characteristics of each edge node in the next monitoring period are predicted based on the behavior cycle of each edge node, and the resource load status of the cloud server in the next monitoring period is predicted based on the load cycle and load time relationship table of the cloud server: when the cloud server load index in the next monitoring period is greater than the load threshold, the resource load status in the next monitoring period is determined to be abnormal; otherwise, the resource load status in the next monitoring period is determined to be normal.

[0027] When the resource load status is abnormal in the next monitoring cycle, a computing load alarm will be sent to the user.

[0028] When the resource load status of the next monitoring period is normal, calculate the sum of the behavioral pattern characteristics of each edge node. If the sum of the behavioral pattern characteristics of each edge node is greater than the summation threshold, set the reserved computing power ratio, and update the allocated computing power of each process in the current monitoring period to the product of the allocated computing power and the reserved computing power ratio.

[0029] According to another aspect of this application, a resource sharing and invocation device based on a cloud platform is provided, comprising:

[0030] The data acquisition unit is used to collect cloud server resource data, uploaded data from each process, and node information in real time.

[0031] The process classification unit is used to classify cloud server processes into cloud processes and externally executable processes.

[0032] The computing power allocation unit is used to allocate initial computing power to each process based on the uploaded data of each process, determine the real-time shared resource status based on the node information of each edge node and the resource data of the cloud server, and set the allocated computing power according to the analysis results, and then execute each process with the allocated computing power.

[0033] The period identification unit is used to establish a table of behavioral pattern characteristics of each edge node and load time relationship of the cloud server within an interval period based on the historical data of each edge node and the task log of the cloud server, and to identify the behavioral period of each edge node and the load period of the cloud server.

[0034] The computing power update unit is used to predict the resource load status of the next monitoring period based on the behavior cycle of each edge node and the load cycle of the cloud server, and then update the allocated computing power of each process in the current monitoring period.

[0035] According to another aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium storing a computer program, wherein the computer program is used to control the electronic device where the computer-readable storage medium is located to execute the aforementioned cloud platform-based resource sharing and invocation method during runtime.

[0036] Compared with existing technologies, the beneficial effects of this invention are as follows: This solution effectively improves the efficiency of resource sharing and invocation by collecting cloud server resource data and data uploaded by each process in real time, classifying cloud server processes based on edge node information, and allocating initial computing power. Based on historical data analysis, a table is established to show the relationship between edge node behavior patterns and cloud server load time, identifying the behavior cycle of each node and the server load cycle. This allows the system to predict the resource load status of the next monitoring cycle, thereby dynamically adjusting the allocated computing power for each process. This method not only optimizes resource utilization and reduces unnecessary computing power waste, but also enhances the system's response speed and service stability, especially performing exceptionally well when dealing with sudden high data volume processing demands. Simultaneously, by accurately predicting future load status, potential risks are warned in advance, ensuring service continuity and reliability, and overall improving the intelligence level and operational efficiency of cloud platform resource sharing and invocation. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is a flowchart illustrating the resource sharing and invocation method based on a cloud platform in this embodiment.

[0039] Figure 2 This is a flowchart illustrating the real-time resource allocation method in this embodiment.

[0040] Figure 3 This is a flowchart illustrating the periodic load analysis method in this embodiment.

[0041] Figure 4 This is a schematic diagram of the structure of the cloud platform-based resource sharing and invocation device provided in this embodiment. Detailed Implementation

[0042] To more clearly illustrate the present invention, the following description, in conjunction with preferred embodiments and accompanying drawings, further explains the invention. Similar components in the drawings are indicated by the same reference numerals. Those skilled in the art should understand that the specific description below is illustrative rather than restrictive and should not be construed as limiting the scope of protection of the present invention.

[0043] It should be noted that although the terms first, second, third, etc., may be used in the embodiments of this application for description, these descriptions should not be limited to these terms. These terms are only used to distinguish the descriptions. For example, without departing from the scope of the embodiments of this application, first can also be referred to as second, and similarly, second can also be referred to as first.

[0044] The acquisition, storage, use, and processing of data in this application all comply with the relevant provisions of national laws and regulations.

[0045] Specifically, the cloud platform-based resource sharing and invocation method described in this application is applied to the invocation and allocation of shared computing resources for sudden high-volume data processing within a cloud server; the cloud server described in this application is the central server of the sharing economy platform, responsible for performing task calculations on each edge node and outputting the calculation results to each edge node; in the above application scenario, for the tasks proposed by each edge node and the uploaded data, the cloud server needs to provide computing power support and data processing for each edge node; each edge node is a terminal device used by users to access the cloud server.

[0046] To apply the above-mentioned application scenarios, this application provides a resource sharing and invocation method based on a cloud platform, the flowchart of which can be found in the document. Figure 1 As shown, it includes:

[0047] Step S101 involves real-time collection of cloud server resource data, uploaded data of each process, and node information. The cloud server resource data includes: peak computing power of CPU and GPU, cache size, maximum number of parallel processes, and cloud server processes. The edge nodes are devices that connect users to the cloud server. The uploaded data of each process includes: uploaded data exchanged between each edge node and the cloud server. The node information includes: memory usage ratio of each edge node and processor utilization rate of each edge node. The collection of cloud server resource data, uploaded data of each process, and node information in this application is a technique well-known to those skilled in the art. This application does not impose specific limitations, as long as the collection requirements are met.

[0048] For example, this application does not specifically limit the processor in the processor utilization of each edge node. In this application, the processor utilization is actually the utilization of the data processing components of each edge node.

[0049] Please continue reading. Figure 1 As shown, the cloud platform-based resource sharing and invocation method further includes:

[0050] Step S102: Classify the cloud server processes to divide them into cloud processes and externally executable processes.

[0051] For example, the cloud process mentioned in this application refers to a process that can only be executed by the cloud server due to database permissions or instruction permissions within the cloud server; the externally executable process refers to a process that can be processed by an external edge node, which can be executed by the edge node simply by sending the instruction data of the process to the edge node; at the same time, this application does not specifically limit the classification process of cloud processes and externally executable processes, as long as the classification requirements are met. In an exemplary embodiment, the data permissions and instruction permissions of the cloud server required by the process can be internally or externally marked, and the process can be monitored to see if it contains internal markings. If it contains internal markings, the process is classified as a cloud tutorial; otherwise, it is marked as an externally executable process.

[0052] Specifically, cloud server processes are categorized into cloud processes and externally executable processes. This categorization helps clarify which tasks must be processed in the cloud and which can be offloaded to edge nodes, improving the flexibility and efficiency of task processing, reducing unnecessary cloud resource consumption, and accelerating response speed.

[0053] Please continue reading. Figure 1 As shown, the cloud platform-based resource sharing and invocation method further includes:

[0054] Step S103: Allocate initial computing power to each process based on the uploaded data of each process, determine the real-time shared resource status based on the node information of each edge node and the resource data of the cloud server, and set the allocated computing power according to the analysis results, and then execute each process with the allocated computing power.

[0055] For details, please refer to Figure 2 The diagram shown is a flowchart illustrating the real-time resource allocation method provided in this application, including:

[0056] Step S301: Allocate initial computing power to each process based on the cloud server process classification results and the uploaded data of each process.

[0057] Specifically, in step S301, the initial computing power allocation process is as follows:

[0058] If the j-th process is a cloud process, the minimum computing power signal of the j-th process in the cloud server is used as the initial computing power of the j-th process;

[0059] If process j is an externally executable process, the uploaded data of each process is compared with the instruction data of the corresponding process. If the uploaded data of process j is greater than or equal to the instruction data of process j, the instruction data of process j is transmitted to the edge node of process j. Then, the minimum computing power signal of process j in the edge node is used as the initial computing power of process j, and the edge node corresponding to process j is used as the running device of process j. If the uploaded data of process j is less than the instruction data of process j, the minimum computing power signal of process j in the cloud server is used as the initial computing power of process j, and the cloud server is used as the running device of process j.

[0060] For example, the minimum computing power signal mentioned in this application is the minimum execution resource table allocated to the process by the operating system within the device where the process resides. The data format of the minimum computing power signal is [memory size, cache size, CPU time, GPU time]. The minimum computing power signal is set in the cloud server and the edge node respectively according to the allocation results of their respective operating systems. This application does not specifically set it.

[0061] Please continue reading. Figure 2 As shown, the real-time resource allocation method further includes:

[0062] Step S302: Obtain historical data of each edge node and task logs of the cloud server, and analyze the computing power redundancy of each process based on the acquisition results; In this application, the historical data of each edge node and task logs of the cloud server can be obtained by program call.

[0063] Specifically, in step S302, the process of calculating the computing power redundancy of each process is as follows:

[0064] The process command is matched with the historical process commands of the edge node where the process is located and the historical process commands of the cloud server, and the average value of the resources scheduled by the successfully matched historical process commands is used as the computing power redundancy of the process.

[0065] For example, this application does not specifically limit the process of "matching process instructions with the historical process instructions of the edge node where the process is located and the historical process instructions of the cloud server". Those skilled in the art can freely set the matching criteria and matching process. In this application, the matching criteria are that the process instructions are the same, but the data carried by the process instructions are not the same, to search and match the historical process instructions of the edge node where the process is located and the historical process instructions of the cloud server.

[0066] Please continue reading. Figure 2 As shown, the real-time resource allocation method further includes:

[0067] Step S303: Determine the real-time shared resource status using cloud server resource data, and set the allocated computing power for each process based on the real-time shared resource status, node information of each edge node, and initial computing power.

[0068] Specifically, in step S303, the process of setting the computing power allocation for each process is as follows:

[0069] The resource load index of the cloud server is determined by the cloud server resource data and set as α, where α = n1 / N + V1 + V2; where n1 is the number of cloud server processes, N is the ratio of the maximum number of parallel processes, V1 is the proportion of cloud server cache occupied, and V2 is the proportion of cloud server memory occupied.

[0070] The resource load index is judged by setting a load judgment threshold to determine the real-time shared resource status: the real-time shared resource status when the resource load index is less than the load judgment threshold is set as reasonable sharing status, and the real-time shared resource status when the resource load index is greater than or equal to the load judgment threshold is set as over-sharing status.

[0071] If the real-time resource sharing state is over-shared and a1×v1+a2×v2 is less than 40%, the running device of process j is updated to the edge node corresponding to process j, and the sum of the computing power redundancy and the initial computing power of process j is used as the allocated computing power of process j; if the real-time resource sharing state is over-shared and a1×v1+a2×v2 is greater than or equal to 40%, the initial computing power of process j is used as the allocated computing power of process j; if the real-time resource sharing state is reasonable sharing, the initial computing power of process j is used as the allocated computing power of process j.

[0072] Where a1 is the memory usage weight of the edge node corresponding to the j-th process, v1 is the memory usage ratio of the edge node corresponding to the j-th process, a2 is the processor utilization of the edge node corresponding to the j-th process, and v2 is the processor utilization of the edge node corresponding to the j-th process.

[0073] Specifically, the resource allocation mechanism has been further refined, especially by analyzing historical data to calculate the computing power redundancy of each process and adjusting the allocation strategy accordingly. This step enables the system to make more accurate resource scheduling decisions based on past performance, effectively avoiding resource waste or shortages.

[0074] For example, this application does not specifically limit the value of the load judgment threshold. Those skilled in the art can set it freely, as long as the value requirement of the load judgment threshold is met. In this application, the resource load index when the runtime of the standard process in the cloud server exceeds 100% of the runtime without load is used as the load judgment threshold. In this application, "the runtime of the standard process in the cloud server exceeds 100% of the runtime without load" is specifically the judgment condition of sluggish process operation.

[0075] For example, in this application, the priority of processes within the cloud server is allocated by the operating system within the cloud server. In this embodiment, the default process is the process after priority sorting. In an exemplary embodiment, this application provides a method for reallocating priorities after the operating system has allocated priorities, the process of which is as follows:

[0076] Obtain the size of fragmented memory in the cloud server's memory and calculate its ratio with the memory usage. Record the result as the data fragmentation ratio. When the data fragmentation ratio exceeds 0.1%, adjust the priority of the processes to treat processes that upload data with the same data structure within a maximum number of parallel processes as adjacent processes.

[0077] Specifically, initial computing power is allocated to each process based on its uploaded data, and the allocation is determined by assessing the real-time shared resource status. This step, by precisely matching process requirements with available resources, not only improves resource utilization but also ensures service stability and efficiency even under high load.

[0078] Please continue reading. Figure 1 As shown, the cloud platform-based resource sharing and invocation method further includes:

[0079] Step S104: Based on the historical data of each edge node and the task logs of the cloud server, establish a table showing the relationship between the behavior pattern characteristics of each edge node and the load time of the cloud server within the interval period, and identify the behavior cycle of each edge node and the load cycle of the cloud server.

[0080] For example, the interval period described in this application is 24 hours.

[0081] Please see Figure 3 As shown, it is a flowchart illustrating the cyclic load analysis method provided in this application, including:

[0082] Step S401: Determine the behavioral pattern characteristics of each edge node based on the historical data of each edge node, and identify the behavioral cycle of each edge node based on the behavioral pattern characteristics of each edge node; the behavioral cycle is the cycle of the behavioral pattern of each edge node uploading data.

[0083] Specifically, the method for establishing the behavioral pattern features of each edge node is as follows:

[0084] The access timestamps of each edge node are divided into time intervals according to a preset interval. The behavior index of the edge node in each time interval is determined based on the number of times the edge node accesses the cloud server and the amount of data processed in the cloud. This index is denoted as xw(i,k). xw(i,k) is set as the product of the number of times the k-th edge node accesses the cloud server and the amount of data processed in the cloud in the i-th time interval. Here, i and k are numeric subscripts, representing the time interval number and the number of edge nodes, respectively.

[0085] Calculate the coefficient of variation (CV(k)) and maximum behavior index (max(k)) of the behavior index of the k-th edge node within the interval period, and set a preset standard to determine the behavior pattern of each edge node: when CV(k) is greater than 40% and max(k) is less than MAX(k), the i-th time interval and the maximum behavior index max(k) are used as the behavior pattern characteristics of the k-th edge node; when CV(k) is less than or equal to 40%, only max(k) is used as the behavior pattern characteristics of the k-th edge node, where MAX(k) is the peak access constant of the k-th edge node;

[0086] The behavioral pattern features of each edge node are input into the autocorrelation function to obtain the behavioral cycle of each edge node.

[0087] For example, in this application, the time interval is divided with a preset interval of 1 hour, and the access timestamps of each edge node are divided into time intervals by judging the subordinate relationship between the access timestamps of each edge node and each time interval. Meanwhile, the autocorrelation function mentioned in this application is the ACF function, which is a prior art well known to those skilled in the art, and will not be described in detail here. At the same time, the value of the peak access constant MAX(k) of the k-th edge node mentioned in this application can be obtained through constraints. In this application, the constraints are the calculation results of the behavior index when the k-th edge node continuously accesses the cloud server until the access delay of the k-th edge node is greater than 2 seconds.

[0088] Please continue reading. Figure 3 As shown, the periodic load analysis method further includes:

[0089] Step S402: Establish a load schedule for the cloud server based on the task logs of the cloud server, and identify the load cycle of the cloud server based on the load schedule.

[0090] Specifically, the process of creating the load-time relationship table is as follows:

[0091] The cloud server load index fz(i) is set by setting the behavior index of edge nodes in each time period interval, and the cloud server load index in each time period interval is used as the load time relationship table.

[0092] Input the load time relationship table of the cloud server into the autocorrelation function to identify the load cycle of the cloud server.

[0093] Specifically, the implementation process of "inputting the load time relationship table of the cloud server into the autocorrelation function to identify the load cycle of the cloud server" in this application is the same as the process of obtaining the behavior cycle of each edge node mentioned above, and will not be described in detail here.

[0094] Specifically, a table showing the relationship between behavioral patterns and cloud server load time is established based on historical data from each edge node, and periodic patterns are identified. This behavior prediction method based on historical data can provide early warnings of potential load peaks, helping administrators take preventative measures and thus maintain stable system operation.

[0095] Please continue reading. Figure 1 As shown, the cloud platform-based resource sharing and invocation method further includes:

[0096] Step S105: The resource load status of the next monitoring period is predicted based on the behavior cycle of each edge node and the load cycle of the cloud server, and then the allocated computing power of each process in the current monitoring period is updated. In this application, the duration of the monitoring period is not specifically limited. Those skilled in the art can set it freely, as long as the duration of the monitoring period is met. In this application, the duration of the monitoring period can be set to 1 hour.

[0097] Specifically, in step S105, the process of updating the allocated computing power of each process within the current monitoring period is as follows:

[0098] The behavior pattern characteristics of each edge node in the next monitoring period are predicted based on the behavior cycle of each edge node, and the resource load status of the cloud server in the next monitoring period is predicted based on the load cycle and load time relationship table of the cloud server: when the cloud server load index in the next monitoring period is greater than the load threshold, the resource load status in the next monitoring period is determined to be abnormal; otherwise, the resource load status in the next monitoring period is determined to be normal.

[0099] When the resource load status is abnormal in the next monitoring cycle, a computing load alarm will be sent to the user.

[0100] When the resource load status of the next monitoring period is normal, calculate the sum of the behavioral pattern characteristics of each edge node. If the sum of the behavioral pattern characteristics of each edge node is greater than the summation threshold, set the reserved computing power ratio, and update the allocated computing power of each process in the current monitoring period to the product of the allocated computing power and the reserved computing power ratio.

[0101] Specifically, the value of the reserved computing power ratio can be set by the formula ln[e-(behavioral pattern characteristics of each edge node-summary threshold) / summary threshold], where e is the natural logarithm.

[0102] It is worth noting that when calculating the "sum of behavioral pattern characteristics of each edge node" in this application, only the time interval is not calculated, and only the maximum behavioral index is calculated.

[0103] For example, in this application, the value of the aggregation threshold can be set as the product of the maximum number of parallel processes continuously occurring in the stress test results within the cloud server and the amount of data continuously uploaded during the test.

[0104] Specifically, this method uses the behavior cycles of edge nodes and the load cycles of cloud servers to predict future resource load conditions and updates the process allocation computing power within the current monitoring period accordingly. This approach can dynamically adapt to changing workloads, optimize resource allocation, reduce the risk of service interruptions caused by sudden traffic surges, and greatly enhance the reliability and adaptability of the system.

[0105] Please see Figure 4 As shown, it is a schematic diagram of the structure of the cloud platform-based resource sharing and invocation transposition provided in this application, including:

[0106] The data acquisition unit is used to collect cloud server resource data, uploaded data from each process, and node information in real time.

[0107] The process classification unit is used to classify cloud server processes into cloud processes and externally executable processes.

[0108] The computing power allocation unit is used to allocate initial computing power to each process based on the uploaded data of each process, determine the real-time shared resource status based on the node information of each edge node and the resource data of the cloud server, and set the allocated computing power according to the analysis results, and then execute each process with the allocated computing power.

[0109] The period identification unit is used to establish a table of behavioral pattern characteristics of each edge node and load time relationship of the cloud server within an interval period based on the historical data of each edge node and the task log of the cloud server, and to identify the behavioral period of each edge node and the load period of the cloud server.

[0110] The computing power update unit is used to predict the resource load status of the next monitoring period based on the behavior cycle of each edge node and the load cycle of the cloud server, and then update the allocated computing power of each process in the current monitoring period.

[0111] The cloud-based resource sharing and invocation device provided in this application can execute the cloud-based resource sharing and invocation method provided in any embodiment of this application, and has the corresponding functional modules and beneficial effects of the execution method.

[0112] This application also provides a computer-readable storage medium, which is a tangible physical storage medium that can store the aforementioned computer program and various types of data used in the program; the physical storage medium includes, but is not limited to, existing physical storage media or combinations thereof, such as random access memory, read-only memory, optical disk, and hard disk.

[0113] Those skilled in the art will understand that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technique for storing information (such as computer-readable programs, data structures, program modules, or other data). Furthermore, it is known to those skilled in the art that communication media typically contain computer-readable programs, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and can include any information delivery medium.

[0114] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for resource sharing and invocation based on a cloud platform, characterized in that, include: The initial computing power is allocated to each process based on the uploaded data of each process. The real-time shared resource status is determined by the node information of each edge node and the resource data of the cloud server, and the allocated computing power is set. Based on the historical data of each edge node and the task logs of the cloud server, a table is established to show the relationship between the behavioral pattern characteristics of each edge node and the load time of the cloud server, and the behavioral cycle of each edge node and the load cycle of the cloud server are identified. The resource load status for the next monitoring period is predicted based on the behavior cycle of each edge node and the load cycle of the cloud server, and then the allocated computing power of each process in the current monitoring period is updated.

2. The resource sharing and invocation method based on a cloud platform according to claim 1, characterized in that, The initial computing power allocation process is as follows: If the j-th process is a cloud process, the minimum computing power signal of the j-th process in the cloud server is used as the initial computing power of the j-th process; If process j is an externally executable process, the uploaded data of each process is compared with the instruction data of the corresponding process. If the uploaded data of process j is greater than or equal to the instruction data of process j, the instruction data of process j is transmitted to the edge node of process j. Then, the minimum computing power signal of process j in the edge node is used as the initial computing power of process j, and the edge node corresponding to process j is used as the running device of process j. If the uploaded data of process j is less than the instruction data of process j, the minimum computing power signal of process j in the cloud server is used as the initial computing power of process j, and the cloud server is used as the running device of process j.

3. The resource sharing and invocation method based on a cloud platform according to claim 2, characterized in that, Acquire historical data from each edge node and task logs from the cloud server, and analyze the computing power redundancy of each process based on the acquisition results; The process of calculating the computing power redundancy of each process is as follows: The process command is matched with the historical process commands of the edge node where the process is located and the historical process commands of the cloud server, and the average value of the resources scheduled by the successfully matched historical process commands is used as the computing power redundancy of the process.

4. The resource sharing and invocation method based on a cloud platform according to claim 3, characterized in that, The process for allocating computing power to each process is as follows: The resource load index of the cloud server is determined by the cloud server resource data and set as α. The resource load index is judged by the set load judgment threshold to determine the real-time shared resource status: the real-time shared resource status when the resource load index is less than the load judgment threshold is set as reasonable sharing status, and the real-time shared resource status when the resource load index is greater than or equal to the load judgment threshold is set as over-sharing status. If the real-time resource sharing state is over-sharing and a1×v1+a2×v2 is less than 40%, the running device of the j-th process is updated to the edge node corresponding to the j-th process, and the sum of the computing power redundancy of the j-th process and the initial computing power is used as the allocated computing power of the j-th process. If the real-time resource sharing state is over-sharing and a1×v1+a2×v2 is greater than or equal to 40%, the initial computing power of the j-th process is used as the allocated computing power of the j-th process. If the real-time resource sharing status is a reasonable sharing status, the initial computing power of the j-th process is used as the allocated computing power of the j-th process; Where a1 is the memory usage weight of the edge node corresponding to the j-th process, v1 is the memory usage ratio of the edge node corresponding to the j-th process, a2 is the processor utilization of the edge node corresponding to the j-th process, and v2 is the processor utilization of the edge node corresponding to the j-th process.

5. The resource sharing and invocation method based on a cloud platform according to claim 4, characterized in that, The behavioral pattern characteristics of each edge node are determined based on historical data, and the behavioral cycle of each edge node is identified based on the behavioral pattern characteristics; the behavioral cycle is the cycle of the data upload behavior pattern of each edge node. Establish a load schedule for the cloud server based on the task logs of the cloud server, and identify the load cycle of the cloud server based on the load schedule.

6. The resource sharing and invocation method based on a cloud platform according to claim 5, characterized in that, The access timestamps of each edge node are divided into time intervals based on a preset interval. The behavior index of the edge node in each time interval is determined based on the number of times the edge node accesses the cloud server and the amount of data processed in the cloud. This index is denoted as xw(i,k). xw(i,k) is defined as the product of the number of times the k-th edge node accesses the cloud server and the amount of data processed in the cloud in the i-th time interval. Here, i and k are numeric subscripts, representing the time interval number and the number of edge nodes, respectively.

7. The resource sharing and invocation method based on a cloud platform according to claim 6, characterized in that, Calculate the coefficient of variation CV(k) and the maximum behavior index max(k) of the behavior index of the k-th edge node within the interval period, and set a preset standard to determine the behavior pattern of each edge node: when CV(k) is greater than 40% and max(k) is less than MAX(k), the i-th time interval and the maximum behavior index max(k) are used as the behavior pattern features of the k-th edge node. When CV(k) is less than or equal to 40%, only max(k) is used as the behavioral pattern feature of the k-th edge node, where MAX(k) is the k-th peak access constant; The behavioral pattern features of each edge node are input into the autocorrelation function to obtain the behavioral cycle of each edge node; The cloud server load index fz(i) is set by setting the behavior index of edge nodes in each time period interval, and the cloud server load index in each time period interval is used as the load time relationship table. Input the load time relationship table of the cloud server into the autocorrelation function to identify the load cycle of the cloud server.

8. The resource sharing and invocation method based on a cloud platform according to claim 7, characterized in that, The behavior pattern characteristics of each edge node in the next monitoring period are predicted based on the behavior cycle of each edge node, and the resource load status of the cloud server in the next monitoring period is predicted based on the load cycle and load time relationship table of the cloud server: when the cloud server load index in the next monitoring period is greater than the load threshold, the resource load status in the next monitoring period is determined to be abnormal; otherwise, the resource load status in the next monitoring period is determined to be normal. When the resource load status is abnormal in the next monitoring cycle, a computing load alarm will be sent to the user. When the resource load status of the next monitoring period is normal, calculate the sum of the behavioral pattern characteristics of each edge node. If the sum of the behavioral pattern characteristics of each edge node is greater than the summation threshold, set the reserved computing power ratio, and update the allocated computing power of each process in the current monitoring period to the product of the allocated computing power and the reserved computing power ratio.

9. A resource sharing and invocation device based on a cloud platform, applied to the resource sharing and invocation method based on a cloud platform as described in any one of claims 1-8, characterized in that, include: The data acquisition unit is used to collect cloud server resource data, uploaded data from each process, and node information in real time. The process classification unit is used to classify cloud server processes into cloud processes and externally executable processes. The computing power allocation unit is used to allocate initial computing power to each process based on the uploaded data of each process, determine the real-time shared resource status based on the node information of each edge node and the resource data of the cloud server, and set the allocated computing power according to the analysis results, and then execute each process with the allocated computing power. The period identification unit is used to establish a table of behavioral pattern characteristics of each edge node and load time relationship of the cloud server within an interval period based on the historical data of each edge node and the task log of the cloud server, and to identify the behavioral period of each edge node and the load period of the cloud server. The computing power update unit is used to predict the resource load status of the next monitoring period based on the behavior cycle of each edge node and the load cycle of the cloud server, and then update the allocated computing power of each process in the current monitoring period.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program is used to control the electronic device on which the computer-readable storage medium is located to execute the resource sharing and invocation method based on the cloud platform as described in any one of claims 1-8 during runtime.

Images