Server idle state recognition method, apparatus and device, and storage medium
By combining server power consumption and performance data, the idle status of cloud servers can be automatically identified, solving the problems of low efficiency and low accuracy in existing technologies. This achieves efficient and accurate idle status identification and saves labor costs.
Patent Information
- Application Number
- PCT/CN2024/080637
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-06-02
- Filing Date
- 2024-03-07
- Publication Date
- 2024-12-05
AI Technical Summary
Existing technologies suffer from low efficiency, low accuracy, and high labor costs in identifying the idle status of cloud servers.
By acquiring the power consumption and performance data of the target server, and combining the performance idle threshold and power consumption idle threshold, the system automatically identifies whether the server is in a low-performance and stable low-power state, thereby determining whether the server is idle.
It improves the efficiency and accuracy of server idle status identification, saves labor costs, and reduces misjudgments of idle status.
Smart Images

Figure CN2024080637_05122024_PF_FP_ABST
Abstract
Description
Server idle state identification method, device, equipment and storage medium
[0001] This application claims priority to the Chinese patent application filed with the China Patent Office on June 2, 2023, with application number 202310651810.2 and invention name “Server Idle State Identification Method, Device, Equipment and Storage Medium”, the entire contents of which are incorporated by reference into this application. Technical Field
[0002] The present application relates to the field of cloud computing technology, and in particular to a method, apparatus, device, and storage medium for identifying a server idle state. Background Art
[0003] With the growing development of cloud technology, cloud servers are becoming increasingly popular. Compared to traditional servers, renting cloud servers can reduce costs to a certain extent. To ensure timely service, cloud servers are usually powered on. However, if cloud servers remain idle for long periods of time while powered on, this waste of resources can be a serious problem. This is especially true in large data centers, where server capacity continues to grow. To ensure the high quality, efficiency, and cost-effectiveness of these servers, it is necessary to identify whether cloud servers are idle.
[0004] Among them, manually checking whether a service program is running on a cloud server to determine whether the server is idle is a common solution. This solution has problems such as low recognition efficiency, low accuracy, and high labor costs.
[0005] Summary of the Invention
[0006] Multiple aspects of the present application provide a server idle state identification method, apparatus, device and storage medium for automatically identifying the server idle state, improving identification efficiency and accuracy, and saving labor costs.
[0007] An embodiment of the present application provides a method for identifying a server idle state, comprising: obtaining target power consumption data and at least one target performance data of a target server in a current statistical period, wherein the at least one target performance data is performance data adapted to the model of the target server; identifying whether the target server is in a low-performance state in the current statistical period based on the at least one target performance data and its corresponding performance idle threshold in the current statistical period; identifying whether the power consumption state of the target server in the current statistical period is in a stable low-power consumption state relative to a historical statistical period based on the target power consumption data and its corresponding power consumption idle threshold in the current statistical period; if the target server is in a low-performance state in the current statistical period and the power consumption state in the current statistical period is in a stable low-power consumption state relative to a historical statistical period, determining that the target server is in an idle state in the current statistical period.
[0008] An embodiment of the present application also provides a server idle state identification device, including: an acquisition module, an identification module and a determination module; the acquisition module is used to obtain target power consumption data and at least one target performance data of a target server in a current statistical period, and the at least one target performance data is performance data adapted to the model of the target server; the identification module is used to identify whether the target server is in a low-performance state in the current statistical period based on the at least one target performance data and its corresponding performance idle threshold in the current statistical period; the identification module is also used to identify whether the power consumption state of the target server in the current statistical period is in a stable low-power consumption state relative to the historical statistical period based on the target power consumption data and its corresponding power consumption idle threshold in the current statistical period; the determination module is used to determine that the target server is in an idle state in the current statistical period if the target server is in a low-performance state in the current statistical period and the power consumption state in the current statistical period is in a stable low-power consumption state relative to the historical statistical period.
[0009] An embodiment of the present application also provides a server idle state identification device, including: a memory and a processor; the memory is used to store a computer program; the processor is coupled to the memory and is used to execute the computer program to implement the steps in the server idle state identification method.
[0010] An embodiment of the present application also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor implements the steps of the server idle state identification method provided in the embodiment of the present application. BRIEF DESCRIPTION OF THE DRAWINGS
[0011] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:
[0012] FIG1 is a flow chart of a method for identifying a server idle state according to an exemplary embodiment of the present application;
[0013] FIG2 is a flow chart of another method for identifying an idle state of a server provided by an exemplary embodiment of the present application;
[0014] FIG3 is a schematic diagram of a process for identifying a stable low power consumption state provided by an exemplary embodiment of the present application;
[0015] FIG4 is a flow chart of a device for identifying an idle state of a server provided by an exemplary embodiment of the present application;
[0016] FIG5 is a flow chart of a server idle state identification device provided by an exemplary embodiment of the present application. DETAILED DESCRIPTION
[0017] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.
[0018] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.
[0019] In response to the problems of low recognition efficiency, low accuracy, and high labor costs in the existing technology, in an embodiment of the present application, the performance status and power consumption status of the target server are respectively identified from the two dimensions of the server's power consumption and performance. Specifically, within each statistical period, the server's performance data and its performance idle threshold are used to determine whether the server is in a low-performance state. The server's power consumption data and its power consumption idle state are used to determine whether the server is in a low-power state. When the target server is in a low-performance state and is in a stable low-power state relative to the historical statistical period, the target server is determined to be in an idle state. This achieves automatic identification of the server's idle state, improves identification efficiency, and saves labor costs. Furthermore, based on the performance data, the stability determination of power consumption is added to reduce the occurrence of misjudgment of the idle state and improve the accuracy of identifying the idle state.
[0020] The technical solutions provided by the embodiments of the present application are described in detail below with reference to the accompanying drawings.
[0021] FIG1 is a flow chart of a method for identifying a server idle state according to an exemplary embodiment of the present application. As shown in FIG1 , the method includes:
[0022] 101. Obtain target power consumption data and at least one target performance data of a target server in a current statistical period, where the at least one target performance data is performance data adapted to a model of the target server;
[0023] 102. Identify whether the target server is in a low performance state in the current statistical period based on at least one target performance data and its corresponding performance idle threshold in the current statistical period;
[0024] 103. Based on the target power consumption data and its corresponding idle power consumption threshold in the current statistical period, identify whether the power consumption state of the target server in the current statistical period is in a stable low power consumption state relative to the historical statistical period;
[0025] 104. If the target server is in a low performance state during the current statistical period and the power consumption state during the current statistical period is in a stable low power consumption state relative to the historical statistical period, it is determined that the target server is in an idle state during the current statistical period.
[0026] In this embodiment, it is possible to identify whether any server is in an idle state. The target server is used as an example for illustration below. The identification of whether the target server is in an idle state is performed for each statistical period. The statistical period can be 24 hours, 48 hours, a week, or a month, and is not limited thereto. The current statistical period is used as an example for description below.
[0027] Among them, the target power consumption data and at least one target performance data of the target server in the current statistical period can be obtained. The target power consumption data can be the global power consumption data of the target server. Further, the global power consumption data can be the sum of the local power consumption data of the main components on the target server, wherein the main components can include but are not limited to: CPU, memory, hard disk, RAID card, GPU card, network card, fan and motherboard, etc. Alternatively, the target power consumption data can be the power consumption data of one or several main components on the target server, which is not limited. In each statistical period, the server counts the target power consumption data once every period of time (such as 0.5s, 1s or 3s), that is, the number of target power consumption data collected in one statistical period is multiple.
[0028] Among them, at least one target performance data may include but is not limited to: one or more data that can characterize the performance of the server, such as CPU load, memory capacity, hard disk read and write, network delay and bandwidth. In each statistical period, the target performance data can be collected once every period of time (such as 1s, 2s or 5s). That is to say, the number of each type of target performance data collected in each statistical period is multiple. At least one target performance data is performance data adapted to the model of the target server. As shown in Figure 2, the model of the target server may include but is not limited to: storage model, GPU model and other models. Other models may be ordinary computer models, such as CPU models. Storage models may include: cache memory, internal memory, auxiliary memory.
[0029] For example, when the target server is a storage model, it is mainly used to store data. The CPU utilization of the server of this storage model is generally at a low level, so the idle state of the server cannot be determined based on the CPU utilization. The main performance data of the storage model is the hard disk read and write data, so the hard disk read and write data can be used as the target performance parameter of the target server of the storage model; for another example, when the target server is a GPU model, the intensive computing part of the target server is implemented by the GPU, while the non-intensive computing part is still implemented by the CPU. Therefore, the target performance data of the target server include: GPU utilization and CPU utilization; for another example, when the target server is an ordinary computing type (such as a CPU model), its CPU utilization is more important, and related data such as its memory utilization and IO utilization need to be considered. Therefore, the target performance data of the target server include: CPU utilization and its load data. The load data is a measure of the current CPU workload and is defined as the average number of threads in the running queue within a specific time interval. An increase in CPU utilization, memory utilization, and IO consumption rate will increase the load.
[0030] In this embodiment, within each statistical period, each target performance data item corresponds to a performance idle threshold, which changes dynamically across different statistical periods. The performance idle threshold corresponding to each target performance data item is a critical value for the target performance data item, used to measure the performance level of the target performance data item and assist in determining the performance status of the target server. Performance status includes a low-performance state and a non-low-performance state. A low-performance state refers to a server state in which the target server's CPU, GPU, memory, or disk I / O, etc., is at a low utilization rate, while a non-low-performance state refers to a server state in which the target performance data item is less than its corresponding performance idle threshold, indicating poor server performance; a non-low-performance state refers to a server state in which the target performance data item is greater than or equal to its corresponding performance idle threshold, indicating good server performance. Based on this, the performance status of the target server within the current statistical period can be identified based on each target performance data item and its performance idle threshold value within the current statistical period, and further determination can be made as to whether the target server's performance status within the current statistical period is in a low-performance state.
[0031] In this embodiment, within each statistical period, the target power consumption data corresponds to a power consumption idle threshold, which changes dynamically between different statistical periods. The power consumption idle threshold corresponding to each target power consumption data is a critical value of the target power consumption data, used to measure the power consumption of the target power consumption data and assist in determining the power consumption state of the target server. The power consumption state includes a low power consumption state and a non-low power consumption state. The low power consumption state refers to a server state in which the target power consumption data of the target server is less than its corresponding power consumption idle threshold, indicating that the server has low power consumption; the non-low power consumption state refers to a server state in which the target power consumption data of the target server is greater than or equal to its corresponding power consumption idle threshold, indicating that the server has high power consumption. For example, the low power consumption state refers to a server state in which the target power consumption data is less than its corresponding power consumption idle threshold, and the non-low power consumption state refers to a server state in which the target power consumption data is greater than or equal to its corresponding power consumption idle threshold. Based on this, it is possible to identify whether the power consumption state of the target server in the current statistical period is in a stable low power consumption state relative to the historical statistical periods based on the target power consumption data and its corresponding power consumption idle threshold in the current statistical period. Among them, a stable low-power state can be understood as the target server being in a low-power state for a period of time. For example, the period of time can be 6 hours, 12 hours, a day or a week, etc. For another example, the period of time can be 2, 3 or more consecutive statistical periods. If the target server is in a low-power state for 2, 3 or more statistical periods, then the target server is considered to be in a stable low-power state. Correspondingly, an unstable low-power state means that it is not in a low-power state for a period of time, but a non-low-power state has occurred for a period of time. Among them, determining whether the target server is in a stable low-power state mainly involves introducing power consumption fluctuation data to reduce misjudgment of idle state identification and improve recognition accuracy.
[0032] In this embodiment, the idle state of the target server is determined by considering both the performance state and the power consumption state of the target server. If both the performance state and the power consumption state meet the requirements, the target server is considered to be in an idle state. Specifically, if the target server is in a low performance state during the current statistical period, and the power consumption state of the target server during the current statistical period is in a stable low power consumption state relative to the historical statistical periods, then the target server is determined to be in an idle state during the current statistical period.
[0033] Optionally, if the target server is in a non-low-performance state during the current statistical period, it is determined that the target server is in a non-idle state during the current statistical period; if the target server is in an unstable low-power state during the current statistical period, or the target server is in a non-low-power state during the current statistical period, it is determined that the target server is in a non-idle state during the current statistical period.
[0034] In an embodiment of the present application, the performance state and power consumption state of the target server are identified from the two dimensions of the server's power consumption and performance. Specifically, within each statistical period, the server's performance data and its performance idle threshold are used to determine whether the server is in a low-performance state. The server's power consumption data and its power consumption idle state are used to determine whether the server is in a low-power state. When the target server is in a low-performance state and is in a stable low-power state relative to the historical statistical period, the target server is determined to be in an idle state. This achieves automatic identification of the server's idle state, improves identification efficiency, and saves labor costs. Furthermore, based on the performance data, a determination of power consumption stability is added to reduce the occurrence of misjudgments of the idle state and improve the accuracy of identifying the idle state.
[0035] In an optional embodiment, the embodiment of the present application can also maintain the idle cycle number of the target server. The idle cycle number indicates the number of cycles that the target server is continuously in an idle state. Each idle cycle is a statistical cycle, for example, it can be 12 hours, 1 day or a week, etc. The number of cycles that the target server is continuously in an idle state can be 0, 1 or more. When it is determined that the target server is in an idle state during the current statistical period, the idle cycle number is updated. For example, when the current idle cycle of the target server is 1 day, the update of the idle cycle number is implemented by adding 1 to the idle cycle number, indicating that the target server has been idle for 2 consecutive statistical periods; when it is determined that the target server is in a non-idle state during the current statistical period, it means that the target server has ended the idle state and re-entered the non-idle state. At this time, the idle cycle number is set to 0. In this way, when the target server re-enters the idle state, it is convenient to restart the counting of the number of cycles that the target server is in an idle state. Among them, the idle cycle number can assist in observing the long-term usage of the server, such as whether it is in an idle state for a long time, and thus facilitate the operator to accurately judge whether the server needs to be optimized.
[0036] In an optional embodiment, it is determined for each server whether it is in an idle state and how many idle cycles the server has had. Specifically, when determining the performance idle threshold, the performance data of the server group is used as the basis for determination, which is beneficial to improving the accuracy of the performance idle threshold. Optionally, when identifying the idle state of each server in the same server group, the performance idle threshold used by each server in the same server group can be the same. In this embodiment, servers are grouped based on dimensions such as business group (BU), model, and product line, that is, servers of the same model within the same BU and belonging to the same product line are classified into the same group to form a server group, but this is not limited to this.
[0037] Based on the above, an implementation method for identifying whether a target server is in a low performance state in a current statistical period according to at least one target performance data and its corresponding performance idle threshold in the current statistical period includes: for each target performance data, determining the server group where the target server is located, obtaining the historical performance data corresponding to each server in the service group in at least one historical statistical period, wherein, in each historical statistical period, there can be one historical performance data corresponding to each time period (such as 1s, 2s or 5s), that is, the number of historical performance data is multiple; based on the historical performance data corresponding to at least one historical statistical period, generating the performance idle threshold corresponding to the target performance data in the current statistical period; comparing at least one performance data to be compared determined according to at least one target performance data with its corresponding performance idle threshold in the current statistical period; for example, since each target There are multiple performance data, that is, target performance data collected at different times in the current statistical period. When compared with the corresponding performance idle threshold, the target performance data with the largest or smallest value can be selected as the performance data to be compared determined by the target performance data in the current statistical period, and the performance data to be compared can be compared with the corresponding performance idle threshold. Alternatively, a weighted average can be performed on multiple target performance data in the current statistical period, and the result of the weighted average can be used as the performance data to be compared determined by the target performance data in the current statistical period, and the performance data to be compared can be compared with the corresponding performance idle threshold. There is no limitation on this. If at least one performance data to be compared is less than the corresponding performance idle threshold, or at least one performance data to be compared that exceeds a certain proportion threshold in the performance data to be compared is less than the corresponding performance idle threshold, it is determined that the target server is in a low performance state in the current statistical period. For example, assuming there are a total of 5 types of performance data to be compared, among which 3 or more types of performance data to be compared are less than the corresponding performance idle threshold, it is determined that the target server is in a low performance state in the current statistical period.
[0038] Optionally, before comparing at least one performance data to be compared with its corresponding performance idle threshold in the current statistical period, a relaxation coefficient for the performance idle threshold may be added. The relaxation coefficient refers to relaxing (e.g., increasing) the performance idle threshold, thereby expanding the performance idle threshold. Accordingly, it is determined whether the target server has been idle for at least one recent historical statistical period. Preferably, the at least one recent historical statistical period is the previous statistical period. If the target server has been idle for at least one recent historical statistical period, it is considered that the probability of the target server being idle in the current statistical period has increased. Therefore, it is necessary to increase the performance idle threshold. The performance idle threshold corresponding to each of the at least one target performance data is increased according to a preset relaxation coefficient, so that even if the target performance data of the target server has a slight improvement, it will be determined to be in a low performance state, and further, it will be determined to be in an idle state. The relaxation coefficient may be 1.05, 1.07, or 1.10, etc. For example, the performance idle threshold is 10, the target server was in an idle state in the previous statistical period, and the relaxation coefficient was 1.10. Then, the performance idle threshold needs to be increased according to the relaxation coefficient. The increased performance idle threshold is 10×1.10=11. Even if the target performance data of the target server increases to 10.9, it will be judged to be in a low performance state, and further, it will be judged to be in an idle state.
[0039] Optionally, depending on the implementation of the target performance data, the implementation method of generating the performance idle threshold corresponding to the target performance data in the current statistical period is also different, which is exemplified below.
[0040] Example A1: The target performance data is implemented as the number of hard disk reads and writes.
[0041] For the number of hard disk reads and writes, the horizontal axis can be the number of hard disk reads and writes, and the vertical axis can be the number of servers. The distribution of the historical hard disk reads and writes of each server in the server group within at least one historical statistical period can be statistically analyzed to obtain multiple normal distribution information. Each normal distribution information represents the corresponding relationship between the number of servers and the number of hard disk reads and writes within at least one historical statistical period. The higher the peak value of the normal distribution, the more servers there are with the hard disk reads and writes corresponding to the peak value; correspondingly, the lower the peak value of the normal distribution, the fewer servers there are with the hard disk reads and writes corresponding to the peak value. In addition, the larger the number of hard disk reads and writes corresponding to the peak value of the normal distribution on the horizontal axis, the more busy the servers are, and the majority of servers are in a non-low-performance state (such as a high-performance state); correspondingly, the smaller the number of hard disk reads and writes corresponding to the peak value of the normal distribution on the horizontal axis, the more idle the servers are, and the majority of servers are in a low-performance state. Based on the distribution information of the historical hard disk reads and writes of each server in the same server group within at least one historical statistical period, the performance idle threshold corresponding to the hard disk reads and writes of the target server in the current statistical period can be determined.
[0042] Specifically, based on the distribution of historical hard disk read and write times of each server in the server group within at least one historical statistical period, at least one normal distribution information is obtained, wherein the horizontal axis of the normal distribution information represents the hard disk read and write times, the vertical axis represents the number of servers corresponding to each hard disk read and write times, and the peak value of the normal distribution information represents the maximum number of servers corresponding to the same hard disk read and write times; based on the minimum peak value in the at least one normal distribution information, a performance idle threshold corresponding to the target performance data within the current statistical period is generated. For example, the hard disk read and write times corresponding to the minimum peak value in the at least one normal distribution information are directly used as the performance idle threshold corresponding to the target performance data within the current statistical period; for another example, the hard disk read and write times corresponding to the minimum peak value in the at least one normal distribution information are added or subtracted by a set read and write times threshold to obtain the performance idle threshold corresponding to the target performance data within the current statistical period. Regardless of how the performance idle threshold corresponding to the hard disk read and write times is determined, in this embodiment, the value of the performance idle threshold is not limited, and can be, for example, 1,000 times, 5,000 times, or 20,000 times, etc., depending on the specific situation.
[0043] Example A2: The target performance data is implemented as CPU utilization or GPU utilization.
[0044] The same performance idle threshold is used for the same server group. If the CPU / GPU utilization of each server in the group is very low, it indicates that the service performance of each server is relatively low. If the CPU / GPU utilization of each server in the group is very high, it indicates that the service performance of each server is very high. Therefore, the quantile method can be used to determine the performance idle threshold for CPU or GPU utilization. The quantile method divides the CPU or GPU utilization into several equal numerical points according to the probability distribution. For example, it can be divided into 2, 4, or 10 equal parts, with the same number of nodes in each part.
[0045] Specifically: in the case where the target performance data is implemented as CPU utilization, the historical CPU utilization of each server in the server group within at least one historical statistical period is divided into intervals using the quantile method to obtain multiple CPU utilization intervals; for example, the above historical CPU utilization is sorted in order from low to high, and the above historical CPU utilization is divided into ten equal parts using the quantile method (such as the decile), that is, 10 CPU utilization intervals are obtained; optionally, before the interval division, the jump points and obvious abnormal points in the CPU utilization can also be eliminated, and the interval division is performed based on the remaining data after elimination; based on the CPU utilization interval with the smallest range and the known utilization experience value, the performance idle threshold corresponding to the target performance data in the current statistical period is generated. Among them, the CPU utilization interval with the smallest range refers to the CPU utilization interval with the smallest utilization, for example, the CPU utilization interval is 5%-7%, 10%-15%, 50%-80%, 80%-95%, and the CPU utilization interval with the smallest range is 5%-7%. For example, a candidate utilization threshold is determined based on the smallest CPU utilization interval. For example, the highest utilization, lowest utilization, or average utilization in the interval is used as the candidate utilization threshold. The smaller of the candidate utilization threshold and the known empirical utilization value is selected as the performance idle threshold corresponding to the CPU utilization. The known empirical utilization value can be 5%, 10%, or 12%.
[0046] When the target performance data is implemented as GPU utilization, the historical GPU utilization of each server in the server group over at least one historical statistical period is divided into intervals using the quantile method to obtain multiple GPU utilization intervals. Based on the GPU utilization interval with the smallest range and a known empirical utilization value, a performance idle threshold corresponding to the target performance data in the current statistical period is generated. The smallest GPU utilization interval refers to the GPU utilization interval with the lowest utilization, for example, GPU utilization intervals of 3%-8%, 11%-16%, 60%-85%, and 86%-96%, where the smallest GPU utilization interval is 3%-8%. For example, based on the smallest GPU utilization interval, a candidate utilization threshold is determined, such as the highest utilization, lowest utilization, or average utilization within the GPU utilization interval. The smaller of the candidate utilization threshold and the known empirical utilization value is selected as the performance idle threshold corresponding to the GPU utilization. The known empirical utilization value can be 7%, 11%, or 12%, for example.
[0047] Example A3: Target performance data is implemented as load data.
[0048] For load data, the average load of a single core can be obtained based on the number of CPUs. The average load of a single core is basically linearly distributed with the CPU utilization. Therefore, the performance idle threshold of the load data can be determined based on the average load of a single core corresponding to a lower CPU utilization.
[0049] Specifically, based on the number of CPU cores of each server in the server group and the historical load data of each server in at least one historical statistical period, the average load of each core of each server is calculated. For example, the total number of CPU cores of each server and the historical load data in at least one historical statistical period are determined, and the average load of each core is the ratio of the historical load data to the total number of CPU cores. Based on the historical CPU utilization and the average load of each core of each server in at least one historical statistical period, the performance idle threshold corresponding to the target performance data in the current statistical period is generated. For example, a statistical method (such as a quantile method) is used to divide the historical CPU utilization into intervals, and the CPU utilization interval with the smallest range is selected. Since there is a linear correspondence between CPU utilization and the average load of each core, according to this linear correspondence, the average load of each core corresponding to the CPU utilization in the CPU utilization interval with the smallest range is obtained. The average of the selected average loads of the core is taken to obtain the idle threshold corresponding to the load data, or the maximum or minimum value is selected from the selected average loads of the core as the idle threshold corresponding to the load data.
[0050] Among them, the above examples A1-A3 select different types of target performance data for servers of different models, and then select different methods for determining performance idle thresholds based on the distribution of different target performance data. Taking into account the individual differences between different target performance data, it is highly flexible and conducive to improving the accuracy of identifying whether the server is in a low-performance state.
[0051] In an optional embodiment, an implementation method for identifying whether the power consumption state of a target server in a current statistical period is in a stable low-power consumption state relative to a historical statistical period based on target power consumption data and its corresponding power consumption idle threshold in the current statistical period includes: generating a power consumption idle threshold in the current statistical period based on the power consumption data of each server in the server group where the target server is located in the current statistical period; for example, the power consumption data in the current statistical period includes power consumption data collected at multiple time points, calculating the mean and standard deviation of the power consumption data collected at multiple time points, and generating a power consumption idle threshold in the current statistical period based on the mean and several standard deviations, such as subtracting N times the standard deviation from the mean to obtain the power consumption idle threshold in the current statistical period, where N is an empirical value and N is a natural number.
[0052] The power consumption data to be compared is determined based on the target power consumption data. For example, the target power consumption data within the current statistical period includes the target power consumption data collected at multiple time points. The power consumption data to be compared may be the maximum or minimum target power consumption data, or the power consumption data to be compared may be the average target power consumption data within the current statistical period. If the power consumption data to be compared determined based on the target power consumption data is less than the power consumption idle threshold, it is determined that the target server is in a low power consumption state within the current statistical period.
[0053] Further, determine whether the target server is in a stable state during the current statistical period. Specifically, obtain the power consumption reference value during the previous historical statistical period. The power consumption reference value during the previous historical statistical period may be the mean or median of the power consumption data during the previous historical statistical period. Determine whether the fluctuation of the target power consumption data relative to the power consumption reference value is within a set fluctuation range. For example, determine whether the absolute value of the difference between the target power consumption data and the power consumption reference value is within a set fluctuation range. The set fluctuation range may be 1-5 watts, 10-15 watts, or 3-17 watts. If the fluctuation of the target power consumption data relative to the power consumption reference value is within the set fluctuation range, determine that the low power consumption state of the target server during the current statistical period is stable relative to the previous historical statistical period. Optionally, if the fluctuation of the target power consumption data relative to the power consumption reference value is not within the set fluctuation range, determine that the low power consumption state of the target server during the current statistical period is unstable relative to the previous historical statistical period, indicating that the power consumption fluctuation of the target server does not meet the requirements.
[0054] Optionally, the method for determining the power consumption reference value in the current statistical period is not limited.
[0055] Wherein, when the power consumption state of the target server in the current statistical period is in a stable low power consumption state relative to the historical statistical period, the power consumption reference value in the current statistical period is determined according to the target power consumption data of the target server in the current statistical period and the power consumption reference value in the previous historical statistical period; the current statistical period includes target power consumption data collected at multiple time points, the multiple target power consumption data are averaged to obtain average power consumption data, and the average power consumption data and the power consumption reference value in the previous historical statistical period are weighted averaged to obtain the power consumption reference value in the current statistical period;
[0056] Among them, when the power consumption state of the target server in the current statistical period is not in a stable low-power consumption state relative to the historical statistical period, the power consumption reference value in the previous statistical period is no longer referenced, and the power consumption reference value in the current statistical period is determined directly based on the target power consumption data of the target server in the current statistical period. For example, after taking the average of the target power consumption data in the current statistical period, the average power consumption data obtained is used as the power consumption reference value in the current statistical period; for another example, the maximum or minimum target power consumption data in the target power consumption data in the current statistical period is used as the power consumption reference value in the current statistical period. Figure 3 shows a schematic diagram of identifying whether the target server is in a stable low-power consumption state and updating the power consumption reference value.
[0057] In an optional embodiment, in order to improve the accuracy of determining whether the target server is in an idle state, abnormal data needs to be eliminated during the process of acquiring the target power consumption data and at least one target performance data.
[0058] Specifically: within the current statistical period, the initial power consumption data and various initial performance data of the target server are collected according to the collection period, and the collection period is less than the statistical period, for example, the collection period can be 1s, 2s or 5s, etc.; the initial power consumption data is abnormally eliminated to obtain the target power consumption data; the various initial performance data are abnormally eliminated to obtain a variety of candidate performance data; for example, when the target server is running, there is a correlation between the target power consumption data and at least one performance data, and the correlation is represented by a polynomial regression model. Accordingly, the target power consumption data and various performance data can be calculated by the polynomial regression model to determine whether the target power consumption data and at least one performance data satisfy the polynomial regression model, thereby eliminating abnormal data.
[0059] Considering the relationship between performance data and power consumption data, performance data that significantly impacts power consumption data is selected as target performance data. However, the performance data that impacts power consumption data varies across different server models. Therefore, at least one target performance data can be selected from multiple candidate performance data sets based on the target server model. Figure 2 illustrates the entire process of acquiring initial power consumption data and multiple initial performance data sets, eliminating anomalies, and determining target performance data based on different server models, as implemented by the data regularization module.
[0060] Optionally, when the target server is a storage model, it is mainly used to store data, so at least the hard disk read and write times are selected as the target performance data from a variety of candidate performance data; when the target server is a GPU computer model, the GPU mainly implements the intensive computing part, and the CPU implements the non-intensive computing part, so at least the GPU utilization and the CPU utilization are selected as the target performance data from a variety of candidate performance data; when the target server is an ordinary computer model, the CPU utilization is more important, and it is also necessary to consider its memory usage, IO consumption and load data and other related data, and select the CPU utilization, IO consumption, memory usage and load data from a variety of candidate performance data as the target performance data, or select the CPU utilization and load data from a variety of candidate performance data as the target performance data. Among them, using different target performance data according to the characteristics of different models avoids the possibility that a single feature cannot accurately identify the idle state, and improves the accuracy of identifying the idle state.
[0061] It should be noted that the execution entity of each step of the method provided in the above embodiment can be the same device, or the method can be executed by different devices. For example, the execution entity of steps 101 to 103 can be device A; for another example, the execution entity of steps 101 and 102 can be device A, and the execution entity of step 103 can be device B; and so on.
[0062] In addition, some of the processes described in the above embodiments and the accompanying drawings include multiple operations that appear in a specific order, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types.
[0063] FIG4 is a schematic diagram of a server idle state identification device provided by an exemplary embodiment of the present application. As shown in FIG4 , the device includes: an acquisition module 41 , an identification module 42 , and a determination module 43 .
[0064] An acquisition module, configured to acquire target power consumption data and at least one target performance data of a target server in a current statistical period, wherein the at least one target performance data is performance data adapted to a model of the target server;
[0065] an identification module, configured to identify whether a target server is in a low performance state during a current statistical period based on at least one target performance data and a corresponding performance idle threshold during the current statistical period;
[0066] The identification module is further configured to identify, based on the target power consumption data and the corresponding power consumption idle threshold value in the current statistical period, whether the power consumption state of the target server in the current statistical period is in a stable low power consumption state relative to the historical statistical period;
[0067] The determination module is used to determine that the target server is in an idle state in the current statistical period if the target server is in a low performance state in the current statistical period and the power consumption state in the current statistical period is in a stable low power consumption state relative to the historical statistical period.
[0068] In an optional embodiment, the device also includes: an update module and a processing module; when it is determined that the target server is in an idle state during the current statistical period, the idle cycle number is updated by the update module, and the idle cycle number represents the number of periods in which the target server is continuously in an idle state; and when it is determined that the target server is in a non-idle state during the current statistical period, the idle cycle number is set to 0 by the processing module.
[0069] In an optional embodiment, the identification module is specifically used to: for each target performance data, generate a performance idle threshold corresponding to the target performance data in the current statistical period based on the historical performance data corresponding to each server in the server group where the target server is located in at least one historical statistical period; compare at least one performance data to be compared determined based on at least one target performance data with its corresponding performance idle threshold in the current statistical period; if at least one performance data to be compared is less than the corresponding performance idle threshold, determine that the target server is in a low performance state in the current statistical period.
[0070] In an optional embodiment, the device also includes a judgment module and an increase module; the judgment module is used to determine whether the target server has been in an idle state in at least one recent historical statistical period; the increase module is used to increase the performance idle threshold corresponding to at least one target performance data according to a preset relaxation coefficient when the target server has been in an idle state in at least one recent historical statistical period.
[0071] In an optional embodiment, the identification module is specifically configured to: when the target performance data is the number of hard disk read and write operations, obtain at least one normal distribution information based on the distribution of the historical number of hard disk read and write operations of each server in the server group within at least one historical statistical period; generate a performance idle threshold corresponding to the target performance data in the current statistical period based on the minimum peak value in the at least one normal distribution information; when the target performance data is CPU utilization or GPU utilization, use a quantile method to divide the historical CPU utilization or historical GPU utilization of each server in the server group within at least one historical statistical period into intervals to obtain multiple CPU utilization intervals or GPU utilization intervals; generate a performance idle threshold corresponding to the target performance data in the current statistical period based on the CPU utilization interval or GPU utilization interval with the smallest range and a known utilization empirical value; when the target performance data is load data, calculate the average load per core of each server based on the number of CPU cores of each server in the server group and the historical load data of each server within at least one historical statistical period; and generate the performance idle threshold corresponding to the target performance data in the current statistical period based on the historical CPU utilization and the average load per core of each server within at least one historical statistical period.
[0072] In an optional embodiment, the identification module is specifically used to: generate a power consumption idle threshold value in the current statistical period based on the power consumption data of each server in the server group where the target server is located in the current statistical period; if the power consumption data to be compared determined based on the target power consumption data is less than the power consumption idle threshold value, determine that the target server is in a low power consumption state in the current statistical period; judge whether the fluctuation of the target power consumption data relative to the power consumption reference value in the previous historical statistical period is within a set fluctuation range; if so, determine that the low power consumption state of the target server in the current statistical period is stable relative to the previous historical statistical period.
[0073] In an optional embodiment, the determination module is used to determine the power consumption reference value in the current statistical period based on the target power consumption data of the target server in the current statistical period and the power consumption reference value in the previous historical statistical period, when the power consumption state of the target server in the current statistical period is in a stable low power consumption state relative to the historical statistical period; and to determine the power consumption reference value in the current statistical period based on the target power consumption data of the target server in the current statistical period, when the power consumption state of the target server in the current statistical period is not in a stable low power consumption state relative to the historical statistical period.
[0074] In an optional embodiment, the acquisition module is specifically used to: collect the initial power consumption data and multiple initial performance data of the target server according to the collection period within the current statistical period, and the collection period is less than the statistical period; eliminate abnormalities in the initial power consumption data to obtain target power consumption data; eliminate abnormalities in the multiple initial performance data to obtain multiple candidate performance data; and select at least one target performance data from the multiple candidate performance data according to the model of the target server.
[0075] In an optional embodiment, the acquisition module is specifically used to: when the target server is a storage model, select at least the number of hard disk read and write times as the target performance data from a plurality of candidate performance data; when the target server is a GPU computer type, select at least GPU utilization and CPU utilization as the target performance data from a plurality of candidate performance data; when the target server is an ordinary computer type, select CPU utilization, IO consumption, memory usage and / or load data as the target performance data from a plurality of candidate performance data.
[0076] The server idle state identification device improved in the embodiment of the present application identifies the performance state and power consumption state of the target server from the two dimensions of power consumption and performance of the server. Specifically, within each statistical period, the server's performance data and its performance idle threshold are used to determine whether the server is in a low-performance state. The server's power consumption data and its power consumption idle state are used to determine whether the server is in a low-power state. When the target server is in a low-performance state and is in a stable low-power state relative to the historical statistical period, the target server is determined to be in an idle state, thereby achieving automatic identification of the server's idle state and saving labor costs. Furthermore, based on the performance data, the stability determination of power consumption is added to reduce the occurrence of idle state misjudgments and improve the accuracy of idle state identification.
[0077] FIG5 is a schematic diagram of a server idle state identification device provided by an exemplary embodiment of the present application. As shown in FIG5 , the device includes a memory 54 and a processor 55 .
[0078] The memory 54 is used to store computer programs and can be configured to store various other data to support operations on the server idle state recognition device. Examples of such data include instructions for any application or method operating on the server idle state recognition device, contact data, phone book data, messages, pictures, videos, etc.
[0079] The memory 54 may be implemented by any type of volatile or non-volatile memory device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.
[0080] The processor 55 is coupled to the memory 54 and is used to execute the computer program in the memory 54, so as to: obtain target power consumption data and at least one target performance data of the target server in the current statistical period, where the at least one target performance data is performance data adapted to the model of the target server; identify whether the target server is in a low-performance state in the current statistical period based on the at least one target performance data and its corresponding performance idle threshold in the current statistical period; identify whether the power consumption state of the target server in the current statistical period is in a stable low-power consumption state relative to the historical statistical period based on the target power consumption data and its corresponding power consumption idle threshold in the current statistical period; if the target server is in a low-performance state in the current statistical period and the power consumption state in the current statistical period is in a stable low-power consumption state relative to the historical statistical period, then it is determined that the target server is in an idle state in the current statistical period.
[0081] In an optional embodiment, the processor 55 is further used to: update the idle cycle number when it is determined that the target server is in an idle state during the current statistical period, where the idle cycle number represents the number of periods in which the target server is continuously in an idle state; and set the idle cycle number to 0 when it is determined that the target server is in a non-idle state during the current statistical period.
[0082] In an optional embodiment, when the processor 55 identifies whether the target server is in a low-performance state in the current statistical period based on at least one target performance data and its corresponding performance idle threshold in the current statistical period, it is specifically used to: for each target performance data, generate a performance idle threshold corresponding to the target performance data in the current statistical period based on the historical performance data corresponding to each server in the server group where the target server is located in at least one historical statistical period; compare at least one performance data to be compared determined based on at least one target performance data with its corresponding performance idle threshold in the current statistical period; if at least one performance data to be compared is less than the corresponding performance idle threshold, determine that the target server is in a low-performance state in the current statistical period.
[0083] In an optional embodiment, before comparing at least one performance data to be compared with its corresponding performance idle threshold in the current statistical period, the processor 55 is also used to: determine whether the target server has been in an idle state in at least one recent historical statistical period; if the target server has been in an idle state in at least one recent historical statistical period, increase the performance idle threshold corresponding to at least one target performance data according to a preset relaxation coefficient.
[0084] In an optional embodiment, when the processor 55 generates the performance idle threshold corresponding to the target performance data in the current statistical period for each target performance data according to the historical performance data corresponding to each server in the server group where the target server is located in at least one historical statistical period, it is specifically used to: when the target performance data is the number of hard disk reads and writes, obtain at least one normal distribution information according to the distribution of the historical hard disk reads and writes of each server in the server group in at least one historical statistical period; generate the performance idle threshold corresponding to the target performance data in the current statistical period according to the minimum peak value in the at least one normal distribution information; when the target performance data is the CPU utilization or GPU utilization, use the quantile method to distribute the performance idle threshold corresponding to each server in the server group. The historical CPU utilization or historical GPU utilization of the server in at least one historical statistical period is divided into intervals to obtain multiple CPU utilization intervals or GPU utilization intervals; based on the CPU utilization interval or GPU utilization interval with the smallest range and the known utilization experience value, a performance idle threshold corresponding to the target performance data in the current statistical period is generated; when the target performance data is load data, the average single-core load of each server is calculated based on the number of CPU cores of each server in the server group and the historical load data of each server in at least one historical statistical period; based on the historical CPU utilization and the average single-core load of each server in at least one historical statistical period, a performance idle threshold corresponding to the target performance data in the current statistical period is generated.
[0085] In an optional embodiment, when the processor 55 identifies whether the power consumption state of the target server in the current statistical period is in a stable low power consumption state relative to the historical statistical period based on the target power consumption data and its corresponding power consumption idle threshold in the current statistical period, it is specifically used to: generate the power consumption idle threshold in the current statistical period based on the power consumption data of each server in the server group where the target server is located in the current statistical period; if the power consumption data to be compared determined based on the target power consumption data is less than the power consumption idle threshold, determine that the target server is in a low power consumption state in the current statistical period; judge whether the fluctuation of the target power consumption data relative to the power consumption reference value in the previous historical statistical period is within a set fluctuation range; if so, determine that the low power consumption state of the target server in the current statistical period is stable relative to the previous historical statistical period.
[0086] In an optional embodiment, the processor 55 is further used to: determine the power consumption reference value in the current statistical period based on the target power consumption data of the target server in the current statistical period and the power consumption reference value in the previous historical statistical period, when the power consumption state of the target server in the current statistical period is in a stable low power consumption state relative to the historical statistical period; determine the power consumption reference value in the current statistical period based on the target power consumption data of the target server in the current statistical period, when the power consumption state of the target server in the current statistical period is not in a stable low power consumption state relative to the historical statistical period.
[0087] In an optional embodiment, when the processor 55 obtains the target power consumption data and at least one target performance data of the target server in the current statistical period, it is specifically used to: collect the initial power consumption data and multiple initial performance data of the target server according to the collection period in the current statistical period, and the collection period is less than the statistical period; eliminate abnormalities in the initial power consumption data to obtain target power consumption data; eliminate abnormalities in multiple initial performance data to obtain multiple candidate performance data; and select at least one target performance data from multiple candidate performance data according to the model of the target server.
[0088] In an optional embodiment, when the processor 55 selects at least one target performance data from a plurality of candidate performance data according to the model of the target server, it is specifically used to: when the target server is a storage model, select at least the hard disk read and write times as the target performance data from a plurality of candidate performance data; when the target server is a GPU computer type, select at least GPU utilization and CPU utilization as the target performance data from a plurality of candidate performance data; when the target server is an ordinary computer type, select CPU utilization, IO consumption, memory usage and / or load data as the target performance data from a plurality of candidate performance data.
[0089] The server idle state identification device improved in the embodiment of the present application identifies the performance state and power consumption state of the target server from the two dimensions of power consumption and performance of the server. Specifically, within each statistical period, the server's performance data and its performance idle threshold are used to determine whether the server is in a low-performance state. The server's power consumption data and its power consumption idle state are used to determine whether the server is in a low-power state. When the target server is in a low-performance state and a stable low-power state, the target server is determined to be in an idle state, thereby achieving automatic identification of the server's idle state and saving labor costs. Furthermore, based on the performance data, the stability determination of power consumption is added to reduce the occurrence of idle state misjudgments and improve the accuracy of idle state identification.
[0090] Furthermore, as shown in FIG5 , the server idle state identification device also includes: a communication component 56, a display 57, a power supply component 58, an audio component 59 and other components. FIG5 only schematically shows some components, which does not mean that the server idle state identification device only includes the components shown in FIG5 . In addition, the components in the dotted box in FIG5 are optional components, not mandatory components, and the specific components may depend on the product form of the working node. The server idle state identification device of this embodiment can be implemented as a terminal device such as a desktop computer, a laptop computer, a smart phone or an IOT device, or as a server-side device such as a conventional server, a cloud server or a server array. If the server idle state identification device of this embodiment is implemented as a terminal device such as a desktop computer, a laptop computer, a smart phone, etc., it may include the components in the dotted box in FIG5 ; if the server idle state identification device of this embodiment is implemented as a server-side device such as a conventional server, a cloud server or a server array, it may not include the components in the dotted box in FIG5 .
[0091] Accordingly, an embodiment of the present application further provides a computer-readable storage medium storing a computer program, which, when executed, can implement the steps in the above method embodiment that can be executed by the method shown in FIG. 1 .
[0092] The above-mentioned memory can be implemented by any type of volatile or non-volatile memory device or a combination thereof, such as static random-access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.
[0093] The above-mentioned communication component is configured to facilitate wired or wireless communication between the device where the communication component is located and other devices. The device where the communication component is located can access a wireless network based on a communication standard, such as WiFi, 2G, 3G, 4G / LTE, 5G and other mobile communication networks, or a combination thereof. In an exemplary embodiment, the communication component receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.
[0094] The above-mentioned display includes a screen, which may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor can not only sense the boundary of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation.
[0095] The power supply assembly provides power to various components of the device in which the power supply assembly is located. The power supply assembly may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the device in which the power supply assembly is located.
[0096] The above-mentioned audio component can be configured to output and / or input audio signals. For example, the audio component includes a microphone (MIC), and when the device where the audio component is located is in an operating mode, such as call mode, recording mode, and voice recognition mode, the microphone is configured to receive external audio signals. The received audio signal can be further stored in a memory or sent via a communication component. In some embodiments, the audio component also includes a speaker for outputting audio signals.
[0097] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to magnetic disk storage, compact disc read-only memory (CD-ROM), optical storage, etc.) that contain computer-usable program code.
[0098] The present application is described with reference to the flow chart and / or block diagram of the method, device (system), and computer program product according to the embodiment of the present application. It should be understood that each flow process and / or box in the flow chart and / or block diagram and the combination of the flow process and / or box in the flow chart and / or block diagram can be realized by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processing machine or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device for realizing the function specified in one flow chart flow or multiple flows and / or one box or multiple boxes of the block diagram.
[0099] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a product including an instruction device that implements the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.
[0100] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.
[0101] In a typical configuration, a computing device includes one or more processors (Central Processing Unit, CPU), input / output interfaces, network interfaces, and memory.
[0102] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.
[0103] Computer-readable media include permanent and non-permanent, removable and non-removable media that can be used to store information using any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change random access memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media such as modulated data signals and carrier waves.
[0104] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.
[0105] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.
Claims
1. A method for identifying a server idle state, characterized in that: include: Acquire target power consumption data and at least one target performance data of a target server in a current statistical period, wherein the at least one target performance data is performance data adapted to a model of the target server; According to the at least one target performance data and its corresponding performance idle threshold in the current statistical period, identifying whether the target server is in a low performance state in the current statistical period; According to the target power consumption data and the power consumption idle threshold value corresponding to it in the current statistical period, identifying whether the power consumption state of the target server in the current statistical period is in a stable low power consumption state relative to the historical statistical period; If the target server is in a low performance state in the current statistical period and the power consumption state in the current statistical period is in a stable low power consumption state relative to the historical statistical period, it is determined that the target server is in an idle state in the current statistical period.
2. The method according to claim 1, characterized in that Also includes: When it is determined that the target server is in an idle state during the current statistical period, updating the idle cycle number, where the idle cycle number indicates the number of cycles during which the target server is continuously in an idle state; as well as When it is determined that the target server is in a non-idle state in the current statistical period, the number of idle periods is set to 0.
3. The method according to claim 1, characterized in that According to the at least one target performance data and a corresponding performance idle threshold value in the current statistical period, identifying whether the target server is in a low performance state in the current statistical period includes: For each type of target performance data, generate a performance idle threshold corresponding to the target performance data in the current statistical period according to the historical performance data corresponding to each server in the server group where the target server is located in at least one historical statistical period; Compare the at least one performance data to be compared determined according to the at least one target performance data with the corresponding performance idle threshold in the current statistical period; If the at least one performance data to be compared is smaller than the corresponding performance idle threshold, it is determined that the target server is in a low performance state in the current statistical period.
4. The method according to claim 3, characterized in that Before comparing the at least one performance data to be compared with the corresponding performance idle threshold in the current statistical period, the method further includes: Determining whether the target server is in an idle state in at least one recent historical statistical period; In the case that the target server is in an idle state in at least one recent historical statistical period, the performance idle threshold corresponding to each of the at least one target performance data is increased according to a preset relaxation coefficient.
5. The method according to claim 3, characterized in that: For each type of target performance data, a performance idle threshold corresponding to the target performance data in the current statistical period is generated according to the historical performance data corresponding to each server in the server group where the target server is located in at least one historical statistical period, including: In the case where the target performance data is the number of hard disk reads and writes, at least one normal distribution information is obtained according to the distribution of the number of historical hard disk reads and writes of each server in the server group in at least one historical statistical period; and according to the minimum peak value in the at least one normal distribution information, a performance idle threshold corresponding to the target performance data in the current statistical period is generated; In the case where the target performance data is CPU utilization or GPU utilization, a quantile method is used to divide the historical CPU utilization or historical GPU utilization of each server in the server group in at least one historical statistical period into intervals to obtain multiple CPU utilization intervals or GPU utilization intervals; based on the CPU utilization interval or GPU utilization interval with the smallest range and the known utilization experience value, a performance idle threshold corresponding to the target performance data in the current statistical period is generated; When the target performance data is load data, the average single-core load of each server is calculated based on the number of CPU cores of each server in the server group and the historical load data of each server in at least one historical statistical period; and the performance idle threshold corresponding to the target performance data in the current statistical period is generated based on the historical CPU utilization and the average single-core load of each server in at least one historical statistical period.
6. The method according to claim 1, characterized in that According to the target power consumption data and the power consumption idle threshold value corresponding to the target server in the current statistical period, identifying whether the power consumption state of the target server in the current statistical period is in a stable low power consumption state relative to the historical statistical period, including: Generate a power consumption idle threshold value in the current statistical period according to the power consumption data of each server in the server group where the target server is located in the current statistical period; If the power consumption data to be compared determined according to the target power consumption data is less than the power consumption idle threshold, it is determined that the target server is in a low power consumption state in the current statistical period; Determine whether the fluctuation of the target power consumption data relative to the power consumption reference value in the previous historical statistical period is within a set fluctuation range; If so, it is determined that the low power consumption state of the target server in the current statistical period is stable relative to the previous historical statistical period.
7. The method according to claim 6, characterized in that Also includes: In a case where the power consumption state of the target server in the current statistical period is in a stable low power consumption state relative to the historical statistical period, determining the power consumption reference value in the current statistical period according to the target power consumption data of the target server in the current statistical period and the power consumption reference value in the previous historical statistical period; When the power consumption state of the target server in the current statistical period is not in a stable low power consumption state relative to the historical statistical period, the power consumption reference value in the current statistical period is determined according to the target power consumption data of the target server in the current statistical period.
8. The method according to any one of claims 1 to 7, characterized in that: Obtain target power consumption data and at least one target performance data of the target server in the current statistical period, including: In the current statistical period, initial power consumption data and various initial performance data of the target server are collected according to a collection period, where the collection period is shorter than the statistical period; Eliminating abnormalities from the initial power consumption data to obtain the target power consumption data; eliminating abnormalities from the multiple initial performance data to obtain multiple candidate performance data; At least one target performance data is selected from the multiple candidate performance data according to the model of the target server.
9. The method according to claim 8, characterized in that According to the model of the target server, at least one target performance data is selected from the multiple candidate performance data, including: In the case where the target server is a storage model, at least hard disk read and write times are selected from the multiple candidate performance data as target performance data; In the case where the target server is a GPU computer type, at least a GPU utilization rate and a CPU utilization rate are selected from the plurality of candidate performance data as target performance data; In the case that the target server is a common computer type, CPU utilization, IO consumption, memory usage and / or load data are selected from the multiple candidate performance data as target performance data.
10. A server idle state identification device, characterized in that: include: Acquisition module, identification module and determination module; The acquisition module is used to acquire target power consumption data and at least one target performance data of the target server in a current statistical period, wherein the at least one target performance data is performance data adapted to the model of the target server; The identification module is used to identify whether the target server is in a low performance state in the current statistical period according to the at least one target performance data and its corresponding performance idle threshold in the current statistical period; The identification module is also used to identify the power consumption state of the target server in the current statistical period relative to the target power consumption data and the corresponding power consumption idle threshold in the current statistical period. Whether it is in a stable low power consumption state during the historical statistical period; The determination module is used to determine that the target server is in an idle state in the current statistical period if the target server is in a low performance state in the current statistical period and the power consumption state in the current statistical period is in a stable low power consumption state relative to the historical statistical period.
11. A server idle state recognition device, characterized in that: include: Memory and processor; The memory is used to store a computer program; the processor is coupled to the memory and is used to execute the computer program to implement the steps in the method according to any one of claims 1 to 9.
12. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the processor is caused to implement the steps in the method according to any one of claims 1 to 9.
Citation Information
Patent Citations
Idle server resource discovery method and device, computer device and storage medium
CN109614231A
Energy consumption management and control device, method and equipment of server
CN112084093A
Execution control method and device, embedded system, equipment and medium
CN112698715A
Abnormal power consumption detection method and device, electronic equipment and storage medium
CN113051127A
Method and device for determining power configuration of server and terminal
CN114764687A
Cited By
Task scheduling method and system, storage medium and computer program product
CN120653406A
