A method and system for evaluating virtual machine migration scheduling in a data center

Abstract

A method and system for evaluating virtual machine migration scheduling in a data center, which relates to the field of electrical digital data processing. In this method, state fingerprint features are constructed; the peak-valley complementarity degree is calculated, and the complementary interval of advantages is identified based on the peak-valley complementarity degree; the potential value of the resource utilization efficiency improvement of the target physical server after migration is calculated; an adaptive scoring baseline is set according to the potential value of the resource utilization efficiency improvement; the matching degree between the resource demand characteristics of the virtual machine to be migrated and the state fingerprint characteristics of the target physical server is compared and analyzed; the migration fitness score is generated by combining the matching degree and the adaptive scoring baseline; and the evaluation result of the virtual machine migration plan to be evaluated is generated according to the migration fitness score and the continuous stability of the complementary interval of advantages. This application is used to improve the accuracy of evaluating virtual machine migration scheduling when new business loads or hardware devices appear in the data center.

Description

Technical Field

[0001] This application belongs to the field of electronic digital data processing, and particularly relates to a method and system for evaluating virtual machine migration scheduling based on a data center. Background Art

[0002] With the continuous expansion of the scale of cloud computing data centers, virtual machine migration scheduling has become an important means of resource optimization management. However, in practical applications, due to the lack of a systematic evaluation system, a large number of virtual machine migration scheduling decisions are often difficult to accurately judge their rationality and effectiveness. This not only affects the formulation of migration plans but may also lead to a large deviation between the post-migration effect and the expectation, resulting in resource waste and performance degradation.

[0003] In related technologies, a quantitative index including multiple evaluation dimensions such as resource utilization rate, load balance degree, service quality, and energy consumption efficiency can be constructed, and an evaluation model can be established in combination with machine learning algorithms to achieve a systematic evaluation of virtual machine migration plans and improve the accuracy of evaluation.

[0004] However, during the actual operation of a data center, since the machine learning model adopted by the existing evaluation method is trained based on historical data, its evaluation accuracy depends on the integrity and representativeness of the training data. When new types of business loads or hardware devices appear in the data center, due to the lack of corresponding historical data support, the evaluation model is difficult to accurately understand and evaluate the migration decision effect in these new scenarios, reducing the accuracy of the evaluation of virtual machine migration scheduling. Summary of the Invention

[0005] This application provides a method and system for evaluating virtual machine migration scheduling based on a data center, which is used to improve the accuracy of evaluating virtual machine migration scheduling when new types of business loads or hardware devices appear in the data center.

[0006] In a first aspect, this application provides a method for evaluating virtual machine migration scheduling based on a data center, which obtains the resource status data of the source physical server and the target physical server included in the virtual machine migration plan to be evaluated before migration;

[0007] Based on the resource status data, a status fingerprint feature of each physical server resource is constructed. The status fingerprint feature is generated by arranging the utilization rates of different resource types in the resource status data in chronological order and extracting key change points;

[0008] Calculate the peak-valley complementarity degree between the source physical server and the target physical server in the resource status data, and identify the complementary interval of resource utilization based on the peak-valley complementarity degree;

[0009] Calculate the potential value of the improvement in resource utilization efficiency of the target physical server after migration according to the complementary interval of advantages and the resource requirement characteristics of the virtual machine to be migrated;

[0010] Set an adaptive scoring baseline according to the potential value of the improvement in resource utilization efficiency;

[0011] Compare and analyze the matching degree between the resource requirement characteristics of the virtual machine to be migrated and the state fingerprint characteristics of the target physical server;

[0012] Generate a migration fitness score by combining the matching degree and the adaptive scoring baseline;

[0013] Generate an evaluation result of the migration plan of the virtual machine to be evaluated according to the migration fitness score and the continuous stability of the complementary interval of advantages.

[0014] By adopting the above technical solution, the resource utilization characteristics of the physical server are characterized by constructing state fingerprint characteristics, and the complementary interval of advantages in resource utilization is identified by combining the peak-valley complementary degree, so that the evaluation plan can accurately grasp the complementarity in resource utilization between the source physical server and the target physical server. Calculate the potential value of the improvement in resource utilization efficiency based on the complementary interval of advantages and set an adaptive scoring baseline, so that the scoring standard can be dynamically adjusted according to the actual effect. By analyzing the matching degree between the virtual machine resource requirement characteristics and the target physical server state fingerprint characteristics, and combining the continuous stability of the complementary interval of advantages, the evaluation result not only reflects the immediate rationality of the migration plan, but also considers the long-term stability of the plan, improving the scientificity and reliability of the migration plan evaluation.

[0015] Combined with some embodiments of the first aspect, in some embodiments, calculate the peak-valley complementary degree between the source physical server and the target physical server in resource state data, and identify the complementary interval of advantages in resource utilization based on the peak-valley complementary degree, specifically including:

[0016] Divide the resource state data into multiple time segments according to a preset time granularity;

[0017] Extract the peaks and valleys of the resource utilization rate data within each time segment to obtain the resource utilization peak periods and resource utilization trough periods of the source physical server and the target physical server;

[0018] Calculate the first overlapping time ratio of the resource utilization peak period of the source physical server and the resource utilization trough period of the target physical server, and the second overlapping time ratio of the resource utilization trough period of the source physical server and the resource utilization peak period of the target physical server;

[0019] Determine the peak-valley complementary degree based on the first overlapping time ratio and the second overlapping time, and mark the time interval with a peak-valley complementary degree greater than a preset threshold as the complementary interval of advantages.

[0020] By adopting the above technical solution, by dividing the resource status data according to a preset time granularity and extracting the peaks and valleys, the periodic characteristics of the resource utilization of the source physical server and the target physical server can be characterized. By calculating the overlapping time ratio of the peak and trough periods of resource utilization to quantify the degree of peak-trough complementarity, the complementarity evaluation becomes more objective and accurate. The method of identifying the complementary interval based on the overlapping time ratio enables the evaluation system to accurately locate the time window with the optimal resource utilization efficiency. This refined analysis method based on time series enables the evaluation system to accurately identify the complementary relationship of resource utilization between physical servers, reducing the performance loss caused by resource competition.

[0021] Combined with some embodiments of the first aspect, in some embodiments, according to the complementary interval and the resource demand characteristics of the virtual machine to be migrated, calculate the potential value of the resource utilization efficiency improvement of the target physical server after migration, specifically including:

[0022] Extract the resource demand fluctuation law of the resource demand characteristics of the virtual machine to be migrated within the complementary interval;

[0023] Based on the resource demand fluctuation law, calculate the resource occupancy of the virtual machine to be migrated on the target physical server at different time points;

[0024] According to the current resource utilization rate and resource occupancy of the target physical server, calculate the resource utilization rate of the target physical server at each time point after migration;

[0025] Based on the change curve of the resource utilization rate of the target physical server before and after migration, calculate the standard deviation of the resource utilization rate within the same time interval;

[0026] Take the ratio of the time interval with a standard deviation less than that before migration after migration as the potential value of the resource utilization efficiency improvement of the target physical server.

[0027] By adopting the above technical solution, by analyzing the resource demand fluctuation law of the virtual machine to be migrated within the complementary interval and calculating its dynamic resource occupancy on the target physical server, the evaluation system can accurately predict the resource utilization situation after migration. Quantifying the potential value of the resource utilization efficiency improvement based on the change of the standard deviation of the resource utilization rate before and after migration enables the evaluation system to objectively measure the actual improvement effect brought by the migration plan. This analysis method based on the volatility of resource utilization not only considers the absolute value change of the resource utilization rate but also pays attention to the improvement of the stability of resource utilization, enabling the evaluation result to more accurately reflect the improvement effect of the migration plan on the overall operation efficiency of the data center.

[0028] Combined with some embodiments of the first aspect, in some embodiments, set an adaptive scoring baseline according to the potential value of the resource utilization efficiency improvement, specifically including:

[0029] Calculate the improvement degree of the target physical server in different resource dimensions based on the potential value of resource utilization efficiency improvement;

[0030] Determine the weight coefficient of each resource dimension according to the current overall load level of the data center;

[0031] Perform weighted calculation on the improvement degree and the weight coefficient to obtain an adaptive scoring baseline.

[0032] By adopting the above technical solution, by calculating the improvement degree of the target physical server in different resource dimensions and determining the weight coefficient according to the current overall load level of the data center, the scoring baseline can adapt to the change of the importance degree of different resource dimensions. This dynamic weight allocation method considering the overall load level of the data center enables the scoring baseline to be adaptively adjusted according to the actual operating environment, reducing the evaluation deviation that may be brought by the fixed scoring standard. The adaptive scoring baseline obtained by performing weighted calculation on the improvement degree and the weight coefficient can more truly reflect the actual value of the migration plan in the current data center environment, improving the practicality of the evaluation result.

[0033] Combined with some embodiments of the first aspect, in some embodiments, generate the evaluation result of the virtual machine migration plan to be evaluated according to the migration fit score and the continuous stability of the complementary advantage interval, specifically including:

[0034] Calculate the duration distribution of the complementary advantage interval in the historical time series;

[0035] Evaluate the stability coefficient of the complementary advantage interval based on the duration distribution;

[0036] Perform non-linear combination operation on the migration fit score and the stability coefficient to obtain the operation result;

[0037] Generate the evaluation result of the virtual machine migration plan to be evaluated according to the operation result.

[0038] By adopting the above technical solution, evaluating the stability coefficient of the complementary advantage interval based on the duration distribution can quantitatively characterize the reliability degree of the resource complementary relationship. Performing non-linear combination operation on the migration fit score and the stability coefficient takes into account both the immediate effect and the long-term stability of the migration plan in the evaluation result, reducing the evaluation deviation caused by relying solely on static matching degree evaluation, making the evaluation result more comprehensive and accurate. By integrating dynamic time series features into the evaluation system, the scientificity and credibility of the virtual machine migration plan evaluation are improved.

[0039] In some embodiments in combination with some embodiments of the first aspect, after the evaluation result of the virtual machine migration plan to be evaluated generated according to the migration fitness score and the continuous stability of the complementary advantage interval, the method further includes:

[0040] Extract the load mutation characteristics of each virtual machine on the target physical server;

[0041] Construct a load mutation propagation chain based on the load mutation characteristics;

[0042] Simulate multiple burst load scenarios based on the load mutation propagation chain;

[0043] Calculate the evaluation scores of the virtual machine migration plan to be evaluated under different burst load scenarios, and extract the stability index of the evaluation scores;

[0044] Correct the evaluation result according to the stability index to obtain the corrected evaluation result.

[0045] By adopting the above technical solution, by extracting the load mutation characteristics of each virtual machine on the target physical server and constructing a load mutation propagation chain, the correlation and influence mode of load fluctuations between virtual machines can be revealed. Simulating multiple burst load scenarios based on the load mutation propagation chain can predict the adaptability of the virtual machine migration plan under different load conditions. By calculating the evaluation scores under different burst load scenarios and extracting the stability index of the scores, the resistance ability of the virtual machine migration plan to load fluctuations can be quantitatively evaluated. Correcting the evaluation result according to the stability index enables the evaluation result to fully consider the impact of burst load on the migration plan, improving the accuracy and reliability of the evaluation result. It can effectively identify the migration plan that still has good effects under load fluctuations, reduce the impact of burst load on the performance of the data center, and improve the stable operation ability of the data center.

[0046] In some embodiments in combination with some embodiments of the first aspect, calculating the evaluation scores of the virtual machine migration plan to be evaluated under different burst load scenarios specifically includes:

[0047] For each burst load scenario, calculate the fluctuation range of each resource dimension in the target physical server;

[0048] Based on the fluctuation range, use the extreme value distribution fitting method to generate the probability density function of the resource utilization rate;

[0049] According to the probability density function, calculate the risk probability that the resource utilization rate after migration exceeds the preset safety threshold;

[0050] Perform a weighted combination of the risk probability and the migration fitness score to obtain the evaluation score.

[0051] By adopting the above technical solution, by calculating the fluctuation range of each resource dimension in the target physical server and using the extreme value distribution fitting method to generate the probability density function of the resource utilization rate, the fluctuation law of resource utilization can be accurately characterized. Based on the probability density function, calculating the risk probability that the resource utilization rate after migration exceeds the preset safety threshold can quantitatively evaluate the resource overrun risk that may be brought by virtual machine migration. Combining the risk probability and the migration fitness score to obtain an evaluation score balances the migration benefits and potential risks in the evaluation result. Modeling and analyzing the resource utilization situation in the scenario of burst load through probability statistics methods provides a more objective and reliable evaluation basis, which helps to select a migration plan that can not only improve the resource utilization efficiency but also ensure the safe and stable operation of the system.

[0052] In a second aspect, an embodiment of the present application provides a virtual machine migration scheduling evaluation system based on a data center, and the virtual machine migration scheduling evaluation system based on the data center includes: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the system to execute the method described in the first aspect and any possible implementation manner in the first aspect.

[0053] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, including instructions, when the above instructions run on the system, enabling the above system to execute the method described in the first aspect and any possible implementation manner in the first aspect.

[0054] In a fourth aspect, an embodiment of the present application provides a computer program product, when the computer program product runs on the system, enabling the system to execute the method described in any possible implementation manner in the first aspect.

[0055] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0056] 1. The present application provides a virtual machine migration scheduling evaluation method based on a data center. By constructing state fingerprint features to characterize the resource utilization characteristics of physical servers and combining the peak-valley complementarity degree to identify the complementary interval of resource utilization, the evaluation scheme can accurately grasp the complementarity of resource utilization between the source physical server and the target physical server. Calculating the potential value of resource utilization efficiency improvement based on the complementary interval and setting an adaptive scoring baseline enables the scoring standard to be dynamically adjusted according to the actual effect. By analyzing the matching degree between the virtual machine resource demand characteristics and the target physical server state fingerprint characteristics and combining the continuous stability of the complementary interval, the evaluation result not only reflects the immediate rationality of the migration plan but also considers the long-term stability of the plan, improving the scientificity and reliability of the migration plan evaluation.

[0057] 2. The present application provides a method for evaluating virtual machine migration scheduling in a data center. By analyzing the resource demand fluctuation law of the virtual machine to be migrated within the complementary interval and calculating its dynamic resource occupancy of the target physical server, the evaluation system can accurately predict the resource utilization status after migration. Quantifying the potential value of resource utilization efficiency improvement based on the standard deviation change of resource utilization rate before and after migration enables the evaluation system to objectively measure the actual improvement effect brought by the migration plan. This analysis method based on resource utilization volatility not only considers the absolute value change of resource utilization rate but also pays attention to the improvement of resource utilization stability, making the evaluation result more accurately reflect the promotion effect of the migration plan on the overall operation efficiency of the data center.

[0058] 3. The present application provides a method for evaluating virtual machine migration scheduling in a data center. By extracting the load mutation characteristics of each virtual machine on the target physical server and constructing a load mutation propagation chain, the correlation and influence mode of load fluctuations between virtual machines can be revealed. Simulating various burst load scenarios based on the load mutation propagation chain can foresee the adaptability of the virtual machine migration plan under different load conditions. By calculating the evaluation scores under different burst load scenarios and extracting the stability index of the scores, the resistance ability of the virtual machine migration plan to load fluctuations can be quantified. Correcting the evaluation result according to the stability index enables the evaluation result to fully consider the impact of burst load on the migration plan, improving the accuracy and reliability of the evaluation result. It can effectively identify the migration plan with good effects under load fluctuations, reduce the impact of burst load on the data center performance, and enhance the stable operation ability of the data center. Description of the Drawings

[0059] Figure 1 is a schematic flowchart of a method for evaluating virtual machine migration scheduling in a data center according to an embodiment of the present application.

[0060] Figure 2 is a schematic flowchart of an optimization method for an evaluation method considering the impact of burst load according to an embodiment of the present application.

[0061] Figure 3 is a schematic structural diagram of an entity device of a virtual machine migration scheduling evaluation system in a data center provided by an embodiment of the present application. Detailed Embodiments

[0062] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the specification and appended claims of the present application, the singular forms "a", "an", "the", "above-mentioned", "said", and "this" are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in the present application refers to any or all possible combinations including one or more of the listed items.

[0063] Hereinafter, the terms "first" and "second" are only used for descriptive purposes and should not be construed as implying or suggesting relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the embodiments of the present application, unless otherwise specified, the meaning of "a plurality" is two or more.

[0064] Next, a description will be given of a method for evaluating virtual machine migration scheduling in an embodiment of the present application by using an example in combination with Figure 1 , as follows:

[0065] Please refer to Figure 1 , which is a flowchart of a method for evaluating virtual machine migration scheduling in an embodiment of the present application.

[0066] S101. Obtain the resource status data of the source physical server and the target physical server included in the virtual machine migration plan to be evaluated before migration;

[0067] In this step, the system needs to obtain the resource status data of the source physical server and the target physical server involved in the virtual machine migration plan before migration execution. These resource status data may include, but are not limited to, the usage conditions of various resources such as CPU utilization, memory utilization, disk I / O utilization, and network bandwidth utilization. The system can collect various resource indicators of the physical server in real time through a resource monitoring component and store the collected data in a time series manner. In addition to real-time collection, the system can also obtain the resource status data of the physical server in a specific time period from the historical monitoring database.

[0068] Specifically, the system can regularly report the resource usage through the resource monitoring deployed on the physical server, and the monitoring period can be set to the second level or the minute level. For historical monitoring data, the system can set a data retention policy and use tools such as a time series database to manage and query and analyze the monitoring data. When obtaining the resource status data, the system can flexibly select the time granularity and data volume of the data according to the evaluation time span.

[0069] S102. Construct the state fingerprint features of each physical server's resources based on the resource status data;

[0070] The system constructs the state fingerprint features of each physical server's resources based on the resource status data. The state fingerprint features are generated by arranging the utilization rates of different resource types in the resource status data in chronological order and extracting the key change points.

[0071] In this step, the system needs to construct the state fingerprint features of the physical server's resources based on the obtained resource status data. The state fingerprint features are an abstract representation of the usage status of the physical server's resources. The purpose is to form a compact and representative feature vector by extracting the key change points of the resource utilization rate and arranging them in chronological order. The system can construct the state fingerprint features for different types of resources respectively, such as the state fingerprint features of the CPU, the state fingerprint features of the memory, etc.

[0072] When specifically implemented, the system can process the resource status data in the form of a sliding window, and each window corresponds to a time segment. Within each time segment, the system analyzes the resource utilization rate data and extracts the key change points therein, such as the peak and valley positions of the utilization rate, or the inflection point positions of the utilization rate change, etc. The extracted key points are arranged in chronological order to form the state fingerprint features of the resource. Considering the volatility of the resource utilization rate, the system can also smooth the original sampling data to eliminate the noise interference on the short-time scale.

[0073] S103. Calculate the peak-valley complementarity degree between the source physical server and the target physical server in the resource status data, and identify the complementary advantage interval of resource utilization based on the peak-valley complementarity degree;

[0074] The system calculates the peak-valley complementarity degree between the source physical server and the target physical server in the resource status data, and identifies the complementary advantage interval of resource utilization based on the peak-valley complementarity degree, which specifically includes: dividing the resource status data into multiple time segments according to a preset time granularity; extracting the peaks and valleys of the resource utilization rate data within each time segment to obtain the resource utilization peak periods and resource utilization valley periods of the source physical server and the target physical server; calculating the first overlapping time ratio of the resource utilization peak period of the source physical server and the resource utilization valley period of the target physical server, and the second overlapping time ratio of the resource utilization valley period of the source physical server and the resource utilization peak period of the target physical server; determining the peak-valley complementarity degree based on the first overlapping time ratio and the second overlapping time, and marking the time interval with the peak-valley complementarity degree greater than the preset threshold as the complementary advantage interval.

[0075] In this step, the system needs to analyze the peak-valley complementarity degree of the source physical server and the target physical server in terms of resource status data, and identify the advantageous complementary intervals for resource utilization. The peak-valley complementarity degree reflects the coincidence of the peak period of resource utilization of the source physical server and the trough period of resource utilization of the target physical server, as well as the coincidence of the trough period of resource utilization of the source physical server and the peak period of resource utilization of the target physical server. The higher the complementarity degree, the more it means that when the resource utilization rate of the source physical server is relatively high, the target physical server has sufficient idle resources to accept the migrated virtual machines, and vice versa.

[0076] To calculate the peak-valley complementarity degree, the system first divides the resource status data into multiple time segments according to a preset time granularity (such as 1 hour, 30 minutes, etc.). For each time segment, the system respectively conducts peak-valley analysis on the resource utilization rate data of the source physical server and the target physical server, and identifies the peak period and trough period of resource utilization by setting a utilization threshold or using statistical methods (such as Kmeans clustering). Then, the system calculates the overlapping proportion in time between the peak period of the source physical server and the trough period of the target physical server, and the overlapping proportion between the trough period of the source physical server and the peak period of the target physical server. The larger the overlapping proportion, the higher the resource complementarity degree between the two physical servers in this time segment.

[0077] Based on the calculated overlapping proportion, the system can set a complementarity degree threshold, and mark the consecutive time segments whose peak-valley complementarity degree exceeds this threshold as the advantageous complementary intervals for resource utilization. These advantageous complementary intervals represent the time periods with the strongest resource complementarity between the source physical server and the target physical server, and are the key objects to be investigated for evaluating the migration performance of virtual machines.

[0078] In practical applications, there may be a situation where the peak and trough periods of the source physical server and the target physical server are misaligned, resulting in the advantageous complementary intervals being refined into multiple discontinuous time periods. To avoid the complementary intervals from being too fragmented, the system can set a minimum complementary interval length threshold. For independent complementary intervals smaller than this threshold, they can be merged with adjacent non-complementary intervals to obtain a more stable and operable complementary interval division result. In addition, the system can also, through manual intervention, have experts fine-tune the boundaries of the complementary intervals according to experience to enhance the rationality of the complementary interval division.

[0079] S104. Calculate the potential value of the resource utilization efficiency improvement of the target physical server after migration according to the advantageous complementary intervals and the resource demand characteristics of the virtual machine to be migrated;

[0080] The system calculates the potential value of improving the resource utilization efficiency of the target physical server after migration according to the complementary interval of advantages and the resource demand characteristics of the virtual machine to be migrated, specifically including: extracting the resource demand fluctuation law of the virtual machine to be migrated within the complementary interval of advantages; calculating the resource occupation of the virtual machine to be migrated on the target physical server at different time points based on the resource demand fluctuation law; calculating the resource utilization rate of the target physical server at each time point after migration according to the current resource utilization rate and resource occupation of the target physical server; calculating the standard deviation of the resource utilization rate within the same time interval based on the change curve of the resource utilization rate of the target physical server before and after migration; taking the proportion of the time interval with a standard deviation less than that before migration after migration as the potential value of improving the resource utilization efficiency of the target physical server.

[0081] In this step, the system needs to predict the potential value of improving the resource utilization efficiency that may be brought about after migrating the virtual machine to the target physical server according to the identified complementary interval of advantages and the resource demand characteristics of the virtual machine to be migrated. The potential value of improving the resource utilization efficiency reflects the improvement space for balancing resource load and reducing resource waste by migrating the virtual machine across servers within the complementary interval of advantages.

[0082] First, the system needs to extract the resource demand fluctuation law of the virtual machine to be migrated within the complementary interval of advantages, including CPU demand, memory demand, disk I / O demand, network bandwidth demand, etc. These resource demand data can be obtained through the analysis of the virtual machine's historical monitoring data, or can be deduced through the prediction model of the virtual machine's business.

[0083] Then, the system matches the resource demand fluctuation law of the virtual machine with the current resource utilization rate of the target physical server, and calculates the additional resource occupation of the target physical server after the virtual machine is migrated at different time points. The additional occupation can be directly superimposed on the original resource utilization rate curve of the target physical server to obtain the expected resource utilization rate level of the target physical server at each time point after migration.

[0084] To quantify the potential of improving the resource utilization efficiency, the system can compare the fluctuation of the resource utilization rate curve of the target physical server before and after migration, and use statistical indicators such as the standard deviation to measure the balance of the resource utilization rate. If the resource utilization rate curve after migration tends to be stable as a whole and the standard deviation decreases, it indicates that migrating the virtual machine helps to alleviate the peak of resource usage of the target physical server, and the resource utilization efficiency is optimized. The system can take the proportion of the time interval with a standard deviation less than that before migration after migration in the entire complementary interval of advantages as a measure of the improvement potential. The higher the proportion, the greater the improvement potential.

[0085] S105. Set an adaptive scoring baseline according to the potential value of improving the resource utilization efficiency;

[0086] The system sets an adaptive scoring baseline according to the potential value of resource utilization efficiency improvement, specifically including: calculating the improvement degree of the target physical server in different resource dimensions based on the potential value of resource utilization efficiency improvement; determining the weight coefficients of each resource dimension according to the current overall load level of the data center; and performing weighted calculation on the improvement degree and the weight coefficients to obtain the adaptive scoring baseline.

[0087] After calculating the potential value of resource utilization efficiency improvement of the target physical server, the system needs to set an adaptive scoring baseline accordingly, which serves as a reference scale for evaluating the pros and cons of the virtual machine cross-server migration plan. Different from a fixed scoring threshold, the adaptive scoring baseline can dynamically adjust the judgment criteria for the improvement degree of resource utilization efficiency according to the change of the overall load level of the data center, showing stronger scenario adaptability.

[0088] In specific implementation, the system first needs to convert the potential value obtained in the previous step into the improvement percentage of each resource dimension, including the improvement percentage of CPU utilization rate, the improvement percentage of memory utilization rate, the improvement percentage of disk I / O utilization rate, the improvement percentage of network bandwidth utilization rate, etc. These improvement percentages represent the optimization effects of virtual machine migration in different resource dimensions.

[0089] Next, the system needs to determine the weight coefficients of each resource dimension to quantify the importance of different resources in comprehensive scoring. An intuitive method for setting weights is to refer to the supply and demand situation of various resources in the data center. For a data center with resource shortages, the weight coefficients of the corresponding resource dimensions can be appropriately increased. For example, if the current CPU resources are very scarce, the system can assign a higher weight to the improvement percentage of CPU utilization rate to highlight its influence in the overall scoring.

[0090] After obtaining the improvement percentage and weight coefficients of each resource dimension, the system calculates a comprehensive improvement score through weighted average and uses it as the adaptive scoring baseline for virtual machine migration. When the potential value of resource utilization efficiency improvement of a certain virtual machine migration plan is higher than the scoring baseline, it can be determined that the migration plan is advisable. The level of the scoring baseline reflects the system's expectation for the performance improvement of virtual machine migration, and its value will be dynamically adjusted according to the load status and business requirements of the data center.

[0091] S106. Compare and analyze the matching degree between the resource demand characteristics of the virtual machine to be migrated and the status fingerprint characteristics of the target physical server;

[0092] During the process of evaluating the virtual machine migration plan, the system needs to deeply analyze the resource matching situation between the virtual machine to be migrated and the target physical server. By comparing the resource requirement characteristics of the virtual machine to be migrated with the status fingerprint characteristics of the target physical server, the system can judge the degree of fit between the two in different resource dimensions, providing a more fine-grained reference basis for the migration decision.

[0093] First of all, the system needs to extract the resource requirement characteristics of the virtual machine to be migrated, including the CPU requirement intensity and its change trend, the memory requirement quantity and its change trend, the disk I / O requirement frequency and its change trend, the network bandwidth requirement fluctuation law, etc. These characteristics can be obtained through statistical analysis of the virtual machine's historical monitoring data, and the obtained requirement characteristics are usually presented in the form of time series.

[0094] Secondly, the system needs to read the status fingerprint characteristics of the target physical server during the corresponding time period and align them with the resource requirement characteristics of the virtual machine. Since the resource status fingerprint extracts the key feature points of the resource utilization change, it can accurately reflect the resource usage law of the physical server.

[0095] After aligning the virtual machine requirement characteristics and the physical server fingerprint characteristics, the system can use similarity measurement algorithms to calculate the matching degree between the two in different resource dimensions. Common similarity measurement algorithms include Euclidean distance, Manhattan distance, cosine similarity, etc. The system can calculate the similarity in the CPU dimension, memory dimension, disk I / O dimension, and network bandwidth dimension respectively to obtain a set of dimension matching scores. The higher the dimension matching score, the closer the resource requirement law of the virtual machine is to the resource usage law of the physical server, and the better the resource matching degree.

[0096] When calculating the matching degree, the system can also consider the correlation between the virtual machine resource requirements and the physical server resource status at different time scales. By introducing a time decay factor, the system can assign smaller weights to historical data that is far from the current moment and larger weights to recent data, so as to more accurately depict the dynamic matching relationship between the virtual machine and the physical server.

[0097] S107. Generate a migration fitness score by combining the matching degree and the adaptive scoring baseline;

[0098] After calculating the resource matching degree between the virtual machine to be migrated and the target physical server, the system needs to further combine the adaptive scoring baseline obtained in the previous steps to generate a comprehensive migration fitness score. The migration fitness score reflects the degree of fit between the virtual machine and the physical server in the resource utilization mode, as well as the overall resource optimization effect that the virtual machine migration may bring. It is a key indicator for evaluating the feasibility of the migration plan.

[0099] In specific implementation, the system first needs to perform weighted aggregation on the scores of each dimension of resource matching degree to obtain a unified matching metric value. When determining the weight coefficients, the system can refer to the importance ranking of different resource dimensions and the current tension levels of various resources in the data center. Usually, the matching degrees of CPU and memory resources are given relatively high weights, while the matching degree weights of disk I / O and network bandwidth resources are relatively low.

[0100] Then, the system compares the aggregated resource matching metric value with the adaptive scoring baseline and generates a migration fitness score through certain mathematical operations. Common operation methods include weighted summation, product summation, etc. For example, the resource matching metric value can be subtracted from the scoring baseline to obtain a relative fitness score, and then it is multiplied by a scaling factor to make the final score fall within the range of 0 to 100 points.

[0101] In the process of synthesizing the migration fitness score, the system can also embed some additional bonus items and deduction items to comprehensively consider various influencing factors of virtual machine migration. For example, if the peak-valley complementarity of the virtual machine and the physical server in terms of CPU and memory resources is very strong, the system can give a certain bonus; on the contrary, if the migration of the virtual machine will lead to increased resource competition among other virtual machines, the system can impose a certain deduction penalty.

[0102] In addition to the above basic score synthesis methods, the system can also adopt some more complex machine learning models to automatically learn the optimal combination method of resource matching degree and scoring baseline by training historical migration case data. This method can fully explore the non-linear correlations between various evaluation indicators and improve the accuracy and generalization of the migration fitness score.

[0103] S108. Generate an evaluation result of the virtual machine migration plan to be evaluated according to the migration fitness score and the continuous stability of the complementary advantage interval.

[0104] The system generates an evaluation result of the virtual machine migration plan to be evaluated according to the migration fitness score and the continuous stability of the complementary advantage interval, specifically including: calculating the duration distribution of the complementary advantage interval in the historical time series; evaluating the stability coefficient of the complementary advantage interval based on the duration distribution; performing a non-linear combination operation on the migration fitness score and the stability coefficient to obtain an operation result; generating an evaluation result of the virtual machine migration plan to be evaluated according to the operation result.

[0105] After generating the migration fitness score, the system also needs to evaluate the continuous stability of the identified complementary intervals to ensure that the migration plan formulated according to the resource complementarity law can remain effective for a period of time. A continuously stable complementary interval means that the resource peak-shaving state between the source physical server and the target physical server can be maintained for a long time, and the virtual machine can continuously benefit from the resource complementarity between the two after migration, improving the overall resource utilization rate.

[0106] To measure the stability of the complementary intervals, the system first needs to trace back the historical monitoring data and examine the occurrence frequency and duration of the complementary intervals over a past period of time. The system can traverse data statistical windows of different time scales, such as days, weeks, months, etc., calculate the start and end time points of the complementary intervals within each window period, and record their durations. After obtaining the duration distribution of the complementary intervals, the system can identify stable complementary intervals based on empirical thresholds or clustering algorithms. For example, a complementary interval with a duration exceeding one day can usually be regarded as a stable interval.

[0107] Based on the identification of stable complementary intervals, the system can further calculate a quantified stability coefficient, representing the confidence level that the complementary intervals can be continuously maintained. A common calculation method is to adopt the idea of exponential decay, assign different weights to complementary intervals of different durations, the longer the duration, the higher the weight, and then normalize the weighted durations to obtain the final stability coefficient. The value range of the coefficient is usually between 0 and 1, and the closer it is to 1, the stronger the stability of the complementary interval.

[0108] After obtaining the stability coefficient of the complementary intervals, the system can combine it with the previous migration fitness score and use a certain mathematical model to generate the final evaluation result of the virtual machine migration plan. Considering the possible non-linear interaction effects between the two indicators of stability and fitness, the system can adopt machine learning models such as polynomial regression and support vector machines to fit the mapping relationship between the two and the evaluation result, rather than simple linear weighting. By training historical migration cases and expert scoring data, the system can automatically learn the optimal combination strategy of the stability coefficient and the fitness score, making the evaluation result closer to the actual migration effect.

[0109] The evaluation result can be a comprehensive score on a 100-point scale, representing the overall quality level of the migration plan to be evaluated, or it can be a multi-level evaluation grade, such as "strongly recommended", "recommended", "considerable", and "not recommended", etc. In addition to presenting the final evaluation result, the system can also generate an evaluation report, detailing the key indicators and their calculation processes of the migration plan in terms of resource matching degree, complementary interval stability, etc., and giving targeted optimization suggestions for the existing deficiencies to help decision-makers more comprehensively examine the feasibility and benefits of the migration plan.

[0110] In the above embodiments, by constructing state fingerprint features to characterize the resource utilization characteristics of physical servers, and combining the degree of peak-valley complementarity to identify the complementary interval of resource utilization, the evaluation scheme can accurately grasp the complementarity of resource utilization between the source physical server and the target physical server. Based on the complementary interval, calculate the potential value of resource utilization efficiency improvement and set an adaptive scoring baseline, so that the scoring standard can be dynamically adjusted according to the actual effect. By analyzing the matching degree between the resource demand characteristics of virtual machines and the state fingerprint characteristics of the target physical server, and combining the continuous stability of the complementary interval, the evaluation result not only reflects the immediate rationality of the migration scheme, but also considers the long-term stability of the scheme, improving the scientificity and reliability of the migration scheme evaluation.

[0111] The basic process of an evaluation method for virtual machine migration scheduling in a data center based on the above application embodiments has been described. This evaluation method evaluates the rationality of the migration scheme by analyzing the resource status of physical servers and the resource demand characteristics of virtual machines. However, in actual application scenarios, the load of the data center usually shows the characteristics of dynamic fluctuations, and sudden loads may have a significant impact on the evaluation results of the migration scheme. Therefore, in order to further improve the accuracy and reliability of the evaluation results, the embodiments of this application also provide an optimization method for the evaluation method considering the impact of sudden loads. The following combines Figure 2 to describe an optimization method for the evaluation method considering the impact of sudden loads in the embodiments of this application: Please refer to Figure 2 which is a schematic flowchart of an optimization method for the evaluation method considering the impact of sudden loads in the embodiments of this application.

[0112] S201. Extract the load mutation characteristics of each virtual machine on the target physical server;

[0113] In this step, the system needs to analyze the load changes of each virtual machine running on the target physical server and identify the mutation characteristics of the load. The load mutation characteristics describe the situation where the resource usage of the virtual machine changes drastically in a short period of time, such as a sharp increase or decrease in CPU utilization, a sharp increase or decrease in memory occupancy, etc. Extracting the load mutation characteristics can help the system predict possible sudden load events and evaluate their impact on the virtual machine migration scheme.

[0114] During specific implementation, the system can collect resource monitoring metric data of virtual machines within a certain time range, such as CPU utilization, memory usage, disk I / O rate, network traffic, etc., and construct multiple time series. Then, the system analyzes each time series using anomaly detection algorithms to identify the mutation points of the monitoring metrics. Common anomaly detection algorithms include statistical process control, wavelet analysis, ARIMA models, etc. The system can select a suitable algorithm according to the characteristics of the data and real-time requirements. The detected mutation points can be characterized by various metrics, such as mutation amplitude, mutation duration, resource usage levels before and after the mutation, etc., to form a load mutation feature vector.

[0115] S202. Construct a load mutation propagation chain based on the load mutation features;

[0116] After extracting the load mutation features of each virtual machine, the system needs to further analyze the correlation between virtual machines and construct a load mutation propagation chain. The load mutation propagation chain depicts how the load mutation of one virtual machine affects other virtual machines, reflecting the transmission and diffusion process of the sudden load among virtual machines. Constructing the load mutation propagation chain helps predict cascading failures and evaluate the impact range of sudden loads.

[0117] During specific implementation, the system first needs to identify the dependency relationship between virtual machines, that is, how the running state of one virtual machine affects other virtual machines. This dependency relationship can be inferred by analyzing the business logic and network communication patterns carried by the virtual machines. For example, if two virtual machines often conduct data interaction or jointly provide services for the same business, then there is a certain dependency relationship between them. When identifying the dependency relationship, the system can use techniques such as association rule mining and causal inference in machine learning to automatically learn the association patterns between virtual machines from a large amount of operation and maintenance logs and network communication records.

[0118] Based on the understanding of the virtual machine dependency relationship, the system can use graph theory methods to construct a load mutation propagation chain. Each virtual machine is regarded as a node in the graph, the dependency relationship as a directed edge, and the load mutation feature as an attribute of the node. If the load mutation of node A affects node B, then a directed edge is connected between the two. By analogy, a directed graph is finally formed, reflecting the propagation path of the load mutation within the entire physical server.

[0119] S203. Simulate multiple sudden load scenarios based on the load mutation propagation chain;

[0120] With the load mutation propagation chain, the system can simulate multiple sudden load scenarios and predict the risks that the virtual machine migration plan may face under different abnormal load conditions. By simulating various extreme situations, the system can comprehensively evaluate the robustness of the migration plan, identify weak links, and provide a reference for optimizing decisions.

[0121] In specific implementation, the system can adopt a graph-based Monte Carlo simulation method. Randomly select one or more nodes from the load mutation propagation chain as the source of the burst load, set the initial load intensity, and then, according to the dependency relationship between nodes, propagate the burst load in the graph according to the probability rules. During the propagation process, the system needs to synchronously update the load intensity attributes of each node until the load intensity is lower than the set threshold or the propagation reaches the boundary of the graph. By repeating the random simulation multiple times, the system can obtain a set of representative burst load scenarios.

[0122] To improve the authenticity of the simulation, the system can refer to the data of historical emergencies and learn the intensity distribution and propagation law of load mutation from actual cases. Using data mining techniques such as clustering and association analysis, the system can discover the typical load patterns of different types of emergencies to guide the setting of simulation parameters. In addition, the system can introduce the time factor and consider the dynamic change trend of node load. A feasible idea is to add a time dimension to the graph model to construct a spatio-temporal propagation chain and simulate the evolution process of burst load in time and space.

[0123] S204. Calculate the evaluation scores of the virtual machine migration scheme to be evaluated under different burst load scenarios, and extract the stability index of the evaluation scores;

[0124] The system calculates the evaluation scores of the virtual machine migration scheme to be evaluated under different burst load scenarios, specifically including: for each burst load scenario, calculate the fluctuation range of each resource dimension in the target physical server; based on the fluctuation range, adopt the extreme value distribution fitting method to generate the probability density function of resource utilization; according to the probability density function, calculate the risk probability that the resource utilization after migration exceeds the preset safety threshold; perform a weighted combination of the risk probability and the migration fitness score to obtain the evaluation score. And extract the stability index of the scoring evaluation score.

[0125] After simulating multiple burst load scenarios, the system needs to evaluate the performance of the virtual machine migration scheme to be evaluated under each scenario and calculate the corresponding evaluation scores. The evaluation scores reflect the ability of the migration scheme to cope with burst loads. The higher the score, the stronger the adaptability and reliability of the scheme. At the same time, the system also needs to analyze the fluctuation of the evaluation scores under different scenarios and extract the stability index of the scores to judge the overall risk level of the migration scheme.

[0126] In specific implementation, the system can adopt the following processing flow: First, for each burst load scenario, the system calculates the utilization fluctuation range of each resource dimension on the target physical server. This requires superimposing the baseline values of resource utilization before and after virtual machine migration on the basis of the simulation results of the load mutation propagation chain. The utilization fluctuation range can be expressed in the form of confidence intervals, maximum and minimum values, etc.

[0127] Then, the system can use the GPD (Generalized Pareto Distribution) model in the extreme value distribution theory to fit the utilization rate distribution of each resource dimension and generate the probability density function of the utilization rate. The GPD model can well characterize the tail distribution characteristics of extreme events and is suitable for describing the impact of bursty loads. Through distribution fitting, the system can obtain the probability of extreme utilization rates exceeding the safety threshold.

[0128] Next, the system can convert the probability of extreme utilization rates into risk metrics, such as risk probability, value at risk, etc. The risk probability represents the likelihood that the resource utilization rate exceeds the threshold under a given bursty load scenario. The value at risk further considers the potential business losses caused by the utilization rate exceeding the threshold. The system can pre-define a risk-loss mapping table to estimate the potential loss value according to different utilization rate levels.

[0129] Finally, the system combines the risk metrics with the migration fitness score obtained in the basic evaluation through weighted combination to form a comprehensive evaluation score. The setting of the weights can be determined according to the system's risk preference and the requirements of the service level agreement (SLA). A system with a high risk preference can assign a greater weight to the risk metrics to avoid service interruptions caused by unexpected events; while a system with a low risk preference can emphasize the migration fitness to pursue the improvement of resource utilization rate. By statistically summarizing the evaluation scores under each bursty load scenario, the system can obtain a multi-dimensional score distribution.

[0130] After obtaining the evaluation score distribution, the system also needs to extract some stability metrics to reflect the reliability of the migration plan in a dynamic environment. Common stability metrics include the mean, variance, coefficient of variation, etc. The mean represents the overall performance level of the plan, while the variance and coefficient of variation characterize the dispersion degree of the scores. The smaller the dispersion degree, the stronger the adaptability of the plan to environmental changes. The system can calculate the stability metrics for multiple migration plans to be evaluated respectively and select the optimal plan by comparing the stabilities of different plans.

[0131] S205. Correct the evaluation results according to the stability metrics to obtain the corrected evaluation results.

[0132] After obtaining the comprehensive evaluation score and stability metrics of the migration plan, the system still needs to perform the last step of correction to balance the credibility of the scores and the risk control requirements. The purpose of the correction is to adjust the evaluation results to make them better reflect the expected performance of the plan in the actual dynamic environment and avoid overly optimistic or pessimistic estimates.

[0133] In specific implementation, the system can design a calibration function that takes the evaluation score and the stability index as inputs and generates a calibrated evaluation result. The calibration function can be in a linear or non-linear form, and the parameters can be set according to historical experience or expert knowledge. A simple linear calibration function is to perform a weighted sum of the evaluation score and the stability index, where the weights reflect the importance the system attaches to stability. The larger the weight of the stability index, the more conservative the calibrated evaluation result will tend to be.

[0134] In addition to linear weighting, the system can also use some more complex non-linear calibration functions, such as exponential functions, logarithmic functions, etc. Non-linear functions can introduce some threshold or gating mechanisms to achieve piecewise calibration. For example, when the stability index is below a certain threshold, the calibration amplitude is small, while when the stability index exceeds this threshold, the calibration amplitude increases rapidly. This piecewise strategy can amplify the impact of the evaluation score when stability is good, and quickly reduce the weight of the evaluation score when stability is poor.

[0135] In practical applications, there may be a situation where the correlation between the evaluation score and the stability index is weak, and a simple weighted combination is difficult to accurately reflect the true value of the solution. To address this issue, the system can adopt some machine learning methods to automatically optimize the structure and parameters of the calibration function by training historical data. For example, a multiple regression model can be used to fit the mapping relationship between the evaluation metrics and the actual effect to obtain a data-driven calibration function. Another example is to use reinforcement learning algorithms to allow the system to autonomously try different calibration strategies in a simulated environment and learn the best calibration method through a reward and punishment mechanism.

[0136] After obtaining the calibrated evaluation result, the system can output a final recommendation for the migration plan selection for the decision maker's reference. The recommended content can include the recommended migration time window, the target physical server, the risk points that need special attention, etc. For solutions with a relatively low calibrated evaluation result, the system can also provide some possible optimization measures, such as adjusting the resource configuration of the virtual machine, increasing the redundant capacity, adjusting the load balancing strategy, etc., to help the user further improve the solution.

[0137] In the above embodiments, by extracting the load mutation characteristics of each virtual machine on the target physical server and constructing a load mutation propagation chain, the correlation and influence mode of load fluctuations between virtual machines can be revealed. Based on the load mutation propagation chain, various burst load scenarios are simulated, and the adaptability of the virtual machine migration scheme under different load conditions can be predicted. By calculating the evaluation scores under different burst load scenarios and extracting the stability indicators of the scores, the resistance of the virtual machine migration scheme to load fluctuations can be quantitatively evaluated. According to the stability indicators, the evaluation results are corrected, so that the evaluation results fully consider the impact of burst load on the migration scheme, improving the accuracy and reliability of the evaluation results. Migration schemes with good effects under load fluctuations can be effectively identified, reducing the impact of burst load on the performance of the data center and enhancing the stable operation ability of the data center.

[0138] The system in the embodiments of the present invention application will be described below from the perspective of hardware processing. Please refer to Figure 3 , which is a schematic structural diagram of an entity device of a virtual machine migration scheduling evaluation system provided by an embodiment of the present application.

[0139] It should be noted that Figure 3 the structure of the system shown is only an example and should not impose any limitations on the functions and usage scope of the embodiments of the present invention.

[0140] As Figure 3 shown, the system includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 302 or the program loaded from the storage part 308 into the random access memory (RAM) 303, such as executing the method in the above embodiments. In the RAM 303, various programs and data required for system operation are also stored. The CPU 301, ROM 302, and RAM 303 are connected to each other through a bus 304. The input / output (I / O) interface 305 is also connected to the bus 304.

[0141] The following components are connected to the I / O interface 305: an input section 306 including a camera, an infrared sensor, etc.; an output section 307 including a liquid crystal display (LCD), a speaker, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as required. A removable medium 311 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc. is mounted on the drive 310 as required so that a computer program read therefrom is installed into the storage section 308 as required.

[0142] Specifically, according to an embodiment of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, an embodiment of the present invention includes a computer program product that includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for performing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 309, and / or installed from the removable medium 311. When the computer program is executed by a central processing unit (CPU) 301, various functions defined in the present invention are executed.

[0143] It should be noted that the computer-readable medium shown in the embodiments of the present invention can be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the computer-readable storage medium can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present invention, a computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable computer program. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above.

[0144] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present invention. Among them, each block in the flowchart or block diagram can represent a module, a program segment, or a part of code, and the above module, program segment, or part of code contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order from that marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram or flowchart, and the combination of blocks in the block diagram or flowchart, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0145] As another aspect, the present invention also provides a computer-readable storage medium, which may be included in the system described in the above embodiments; or may exist alone without being assembled into the system. The above storage medium carries one or more computer programs, and when the above one or more computer programs are executed by a processor of a system, the system implements the method provided in the above embodiments.

[0146] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the various embodiments of the present application.

[0147] As used in the above embodiments, depending on the context, the term "when..." can be interpreted as "if...", or "after...", or "in response to determining...", or "in response to detecting...". Similarly, depending on the context, the phrase "when determining..." or "if detecting (the stated condition or event)" can be interpreted as "if determining...", or "in response to determining...", or "when detecting (the stated condition or event)", or "in response to detecting (the stated condition or event)".

[0148] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website, a computer, a server, or a data center to another website, a computer, a server, or a data center by wire (such as coaxial cable, optical fiber, digital subscriber line) or wirelessly (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer, or a data storage device such as a server or a data center that includes one or more integrated available media. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid-state drive), etc.

[0149] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by relevant hardware instructed by a computer program. This program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above method embodiments. The foregoing storage media include: various media such as ROM, random access memory (RAM), magnetic disks, or optical discs that can store program codes.

Claims

1. A data center virtual machine migration scheduling evaluation method, characterized in that: include: Obtain resource status data of a source physical server and a target physical server before migration included in the virtual machine migration solution to be evaluated; Constructing a status fingerprint feature of each physical server resource based on the resource status data, wherein the status fingerprint feature is generated by arranging the utilization rates of different resource types in the resource status data in chronological order and extracting key change points; Calculating the peak-valley complementarity degree between the source physical server and the target physical server on the resource status data, and identifying the advantageous complementary interval of resource utilization based on the peak-valley complementarity degree; Calculating the resource utilization efficiency improvement potential value of the target physical server after migration based on the complementary advantage interval and the resource demand characteristics of the virtual machine to be migrated; Setting an adaptive scoring baseline according to the resource utilization efficiency improvement potential value; Comparative analysis of the matching degree between the resource demand characteristics of the virtual machine to be migrated and the state fingerprint characteristics of the target physical server; generating a migration fit score by combining the matching degree and the adaptive scoring baseline; An evaluation result of the virtual machine migration solution to be evaluated is generated according to the migration compatibility score and the continuous stability of the complementary advantage interval.

2. The method according to claim 1, characterized in that The calculating the peak-valley complementarity degree between the source physical server and the target physical server in the resource status data, and identifying the advantageous complementary interval of resource utilization based on the peak-valley complementarity degree, specifically includes: Dividing the resource status data into multiple time segments according to a preset time granularity; Extracting peaks and troughs of resource utilization data in each time segment to obtain peak periods and trough periods of resource utilization of the source physical server and the target physical server; Calculate a first overlapping time ratio between a resource utilization peak period of the source physical server and a resource utilization valley period of the target physical server, and a second overlapping time ratio between a resource utilization valley period of the source physical server and a resource utilization peak period of the target physical server; The peak-valley complementarity degree is determined based on the first overlapping time ratio and the second overlapping time, and a time interval in which the peak-valley complementarity degree is greater than a preset threshold is marked as an advantageous complementary interval.

3. The method according to claim 1, characterized in that The calculating, according to the complementary advantage interval and the resource demand characteristics of the virtual machine to be migrated, the resource utilization efficiency improvement potential value of the target physical server after the migration specifically includes: Extracting resource demand characteristics of the virtual machine to be migrated and resource demand fluctuation patterns within the complementary advantage interval; Calculate the resource occupation of the target physical server by the to-be-migrated virtual machine at different time points based on the resource demand fluctuation rule; Calculating the resource utilization of the target physical server at each time point after the migration according to the current resource utilization of the target physical server and the resource occupancy; Based on the change curve of the resource utilization of the target physical server before and after the migration, calculating the standard deviation of the resource utilization within the same time interval; The proportion of time intervals in which the standard deviation after migration is less than the standard deviation before migration is used as the potential value for improving the resource utilization efficiency of the target physical server.

4. The method according to claim 1, characterized in that The step of setting an adaptive scoring baseline according to the resource utilization efficiency improvement potential value specifically includes: Calculating the improvement degree of the target physical server in different resource dimensions based on the resource utilization efficiency improvement potential value; Determine the weight coefficient of each resource dimension according to the current overall load level of the data center; The improvement degree and the weight coefficient are weighted and calculated to obtain an adaptive scoring baseline.

5. The method according to claim 1, characterized in that Generating the evaluation result of the virtual machine migration scheme to be evaluated according to the migration compatibility score and the continuous stability of the complementary advantage interval specifically includes: Calculate the duration distribution of the complementary advantage interval in the historical time series; Evaluating the stability coefficient of the complementary advantage interval based on the duration distribution; Performing a nonlinear combination operation on the migration fit score and the stability coefficient to obtain an operation result; An evaluation result of the virtual machine migration solution to be evaluated is generated according to the calculation result.

6. The method according to claim 1, characterized in that After the evaluation result of the virtual machine migration solution to be evaluated is generated according to the migration compatibility score and the continuous stability of the complementary advantage interval, the method further includes: Extracting load mutation characteristics of each virtual machine on the target physical server; Constructing a load mutation propagation chain according to the load mutation characteristics; Simulating various burst load scenarios based on the load mutation propagation chain; Calculating the evaluation score of the virtual machine migration solution to be evaluated under different burst load scenarios, and extracting the stability index of the evaluation score; The evaluation result is corrected according to the stability index to obtain a corrected evaluation result.

7. The method according to claim 6, characterized in that The calculating the evaluation score of the virtual machine migration solution to be evaluated under different burst load scenarios specifically includes: For each of the burst load scenarios, calculating the fluctuation range of each resource dimension in the target physical server; Based on the fluctuation range, an extreme value distribution fitting method is used to generate a probability density function of resource utilization; Calculate the risk probability of resource utilization exceeding a preset safety threshold after migration according to the probability density function; The risk probability and the migration compatibility score are weighted and combined to obtain an evaluation score.

8. A data center virtual machine migration scheduling evaluation system, characterized in that: The system comprises: One or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to cause the system to execute the method as described in any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on a system, the system is caused to execute the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that When the computer program product is run on a system, the system is caused to execute the method according to any one of claims 1 to 7.

Images