Server power control method and device and electronic equipment
By dynamically adjusting server power, combining rated power and priority, the conflict between rack space utilization and power safety is resolved, achieving efficient resource management.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-28
- Publication Date
- 2026-03-10
AI Technical Summary
In existing technologies, rack space utilization is low and power safety is difficult to balance, resulting in wasted rack space.
By dynamically adjusting server power at predicted times, based on rated power, server priority, and adjustment factors, the maximum power usage of servers is limited, ensuring the power safety of the rack in the future.
This approach ensures full utilization of rack space while guaranteeing rack power safety, avoiding space waste, and prioritizing the performance of high-priority servers.
Smart Images
Figure CN121635618A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of communication technology, in particular to a server power control method and device and electronic equipment. BACKGROUND
[0002] In response to the national green low-carbon development policy, the construction of green data centers has become an important direction for the development of the industry. For example, by setting a rated power for each rack in the computer room in advance during the construction phase of the data center, the working power of the rack can be controlled to avoid excessive consumption of power resources.
[0003] Currently, in actual applications, the number of servers to be mounted on a rack is usually planned according to the rated power set for the rack. For example, assuming that the rated power set for a rack is 10 kilowatts (KW), and the rated power of a single server to be mounted is 1 KW, to ensure the power safety of the rack, that is, the sum of the use powers of the servers mounted on the rack at any time does not exceed the rated power set for the rack, at most 10 servers can be mounted on the rack. However, this rack planning method can cause waste of rack space. For example, assuming that the rack space of the rack is 42 units (U), and the rack space required by a single server to be mounted is 2 U, the rack space occupied by 10 servers is 20 U, so the rack space utilization rate of the rack is only about 50%. Therefore, how to ensure that the rack space is fully utilized while ensuring the power safety of the rack is a technical problem to be solved at present. SUMMARY
[0004] Therefore, the present application provides a server power control method and device and electronic equipment to ensure that the rack space is fully utilized while ensuring the power safety of the rack.
[0005] The present application provides a server power control method, which comprises: For each predicted time, if it is found that the sum of the predicted powers of the servers on the target rack at the predicted time is greater than the rated power configured for the target rack, then: For each server, based on the rated power, the predicted power of each server at the predicted time, the priority of the server, and the adjustment factor matched with the priority of the server at the predicted time, the predicted power of the server at the predicted time is adjusted to obtain an adjusted power. The adjustment factor indicates the power adjustment amplitude corresponding to the matched priority. At the same predicted time, the power adjustment amplitude corresponding to a low priority is greater than the power adjustment amplitude corresponding to a high priority, and the greater the adjustment factor, the greater the difference between the power adjustment amplitude corresponding to a low priority and the power adjustment amplitude corresponding to a high priority. based on the adjusted power of each server, determine maximum usage power of each server in a time period from the prediction time to a next prediction time, the maximum usage power of any server limits power of the server to be less than the maximum usage power of the server in the time period from the prediction time to the next prediction time after the prediction time arrives.
[0006] The embodiment of the present application also provides a server power control device, which comprises: The control module is used for, for each prediction time, if it is found that the sum of the predicted powers of the servers on the target rack at the prediction time is greater than the rated power configured for the target rack, then: for each server, based on the rated power, the predicted power of each server at the prediction time, the priority of the server and the adjustment factor matched with the priority of the server at the prediction time, adjust the predicted power of the server at the prediction time to obtain an adjusted power; the adjustment factor indicates a power adjustment amplitude corresponding to the matched priority; at the same prediction time, the power adjustment amplitude corresponding to a low priority is greater than the power adjustment amplitude corresponding to a high priority, and the greater the adjustment factor is, the greater the difference between the power adjustment amplitude corresponding to the low priority and the power adjustment amplitude corresponding to the high priority is; based on the adjusted power of each server, determine maximum usage power of each server in a time period from the prediction time to a next prediction time, the maximum usage power of any server limits power of the server to be less than the maximum usage power of the server in the time period from the prediction time to the next prediction time after the prediction time arrives.
[0007] The embodiment of the present application also provides an electronic device, which comprises: a processor; and a computer readable storage medium, in which computer program instructions are stored, and the computer program instructions make the processor execute steps of the above method when the computer program instructions are run by the processor.
[0008] The embodiment of the present application also provides a computer readable storage medium, in which computer program instructions are stored, and the computer program instructions make the processor execute steps of the above method when the computer program instructions are run by the processor.
[0009] As can be seen from the above technical solutions, in this embodiment, when it is found that the sum of the predicted power of each server on the target rack at each prediction time is greater than the rated power configured for the target rack, i.e., there is a power safety issue, the predicted power of each server at the prediction time is dynamically adjusted based on the rated power, the predicted power of each server at the prediction time, the priority of each server, and the adjustment factor matched with the priority of each server at the prediction time. Based on the adjusted power of each server, the maximum power used by each server in the time period from the prediction time to the next prediction time is determined. The maximum power used limits the power of the server to be less than the maximum power used by the server in the time period from the arrival of the prediction time to the next prediction time, so as to ensure the power safety of the target rack in the time period from the prediction time to the next prediction time. Based on this, by dynamically adjusting the maximum power used by each server on the target rack in the future time period, this embodiment can flexibly plan the number of servers to be racked according to the rack space of the target rack, which effectively avoids the problem of rack space waste, thereby ensuring the full utilization of rack space while ensuring the power safety of the rack.
[0010] Furthermore, in this embodiment, the predicted power of the server is dynamically adjusted by introducing an adjustment factor and the server priority. The adjustment factor indicates the power adjustment range corresponding to the matched priority. At the same prediction time, the power adjustment range corresponding to the lower priority is greater than the power adjustment range corresponding to the higher priority. Moreover, the larger the adjustment factor, the greater the difference between the power adjustment range corresponding to the lower priority and the power adjustment range corresponding to the higher priority. In this way, by suppressing the maximum power used by the lower priority server, the operating performance of the higher priority server can be guaranteed to the greatest extent. Attached Figure Description
[0011] The accompanying drawings, which are incorporated in and form part of this application, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.
[0012] Figure 1 This is a schematic diagram of the method flow provided in the embodiments of this application.
[0013] Figure 2 This is a schematic diagram of another method flow provided for an embodiment of this application.
[0014] Figure 3 This is a schematic diagram of another method flow provided in an embodiment of this application.
[0015] Figure 4 This is a schematic diagram of the device structure provided in the embodiments of this application.
[0016] Figure 5This is a schematic diagram of the electronic device structure provided in an embodiment of this application. Detailed Implementation
[0017] To enable those skilled in the art to better understand the technical solutions provided in the embodiments of this application, and to make the above-mentioned objectives, features and advantages of the embodiments of this application more apparent and understandable, the technical solutions in the embodiments of this application will be further described in detail below with reference to the accompanying drawings.
[0018] See Figure 1 , Figure 1 This is a flowchart illustrating a server power control method provided in an embodiment of this application. In this embodiment, as one example, the method can be applied to an electronic device, such as an operation and maintenance server. Figure 1 As shown, the process may include the following steps: Step 101: For each prediction time, if the sum of the predicted power of each server on the target rack is found to be greater than the rated power configured for the target rack, then for each server, based on the rated power, the predicted power of each server at the prediction time, the priority of the server, and the adjustment factor matched with the priority of the server at the prediction time, the predicted power of the server at the prediction time is adjusted to obtain the adjusted power.
[0019] Step 102: Based on the adjusted power of each server, determine the maximum power used by each server during the period from the current prediction time to the next prediction time. The maximum power used by any server limits the power of that server to be less than its maximum power used during the period from the arrival of the current prediction time to the next prediction time.
[0020] In this embodiment, as an example, the predicted power of each server on the target rack can be predicted periodically (e.g., every 30 minutes or 1 hour) at each prediction time within a future set time period. The specific method of prediction will be described with examples below and will not be elaborated upon here.
[0021] Based on this, for each prediction time, if the sum of the predicted power of all servers on the target rack is found to be greater than the rated power configured for the target rack, then to ensure the power safety of the target rack at that future time, the predicted power of each server at that prediction time can be dynamically adjusted. Specifically, for each server, the predicted power at that prediction time can be adjusted based on the rated power, the predicted power of each server at that prediction time, the server's priority, and the adjustment factor matched with the server's priority at that prediction time, to obtain the adjusted power. Then, based on the adjusted power of each server, the maximum power used by each server during the period from the prediction time to the next prediction time can be determined. The maximum power used by any server limits the power of that server to be less than its maximum power during the period from the arrival of the prediction time to the next prediction time, thereby ensuring that the sum of the power used by all servers at any time during the period from the arrival of the prediction time to the next prediction time will not exceed the rated power of the target rack, thus ensuring the power safety of the target rack during this period.
[0022] If the sum of the predicted power of each server on the target rack at the predicted time is less than or equal to the rated power configured for the target rack, the maximum power used by each server during the period from the predicted time to the next predicted time can be determined directly based on the predicted power of each server at the predicted time.
[0023] In this embodiment, by dynamically adjusting the maximum power consumption of each server on the rack over a future time period, the number of servers to be mounted on the rack can be planned without being limited by the rated power of the servers. Based on this, this embodiment can flexibly plan the number of servers to be mounted according to the rack space, thereby effectively avoiding the problem of rack space waste, and thus ensuring both rack space utilization and rack power safety.
[0024] In this embodiment, the aforementioned adjustment factor indicates the power adjustment magnitude corresponding to the matched priority. At the same prediction time, the power adjustment magnitude corresponding to lower priority is greater than that corresponding to higher priority (i.e., the lower the priority, the larger the power adjustment magnitude), and the larger the adjustment factor, the greater the difference between the power adjustment magnitudes corresponding to lower and higher priority. This embodiment dynamically adjusts the predicted power of the server by introducing an adjustment factor and the server's priority. This can suppress the maximum power usage of lower priority servers, thereby maximizing the performance of higher priority servers while effectively controlling the power consumption of lower priority servers.
[0025] As for how to adjust the predicted power of the server at the predicted time, and how to determine the maximum power used by each server in the time period from the predicted time to the next predicted time based on the adjusted power of each server, examples will be given below, and will not be repeated here.
[0026] This concludes the process. Figure 1 The method flow is shown.
[0027] pass Figure 1 As can be seen from the method flow shown, in this embodiment, when it is found that the sum of the predicted power of each server on the target rack at each prediction time is greater than the rated power configured for the target rack, i.e., there is a power safety issue, the predicted power of each server at the prediction time is dynamically adjusted based on the rated power, the predicted power of each server at the prediction time, the priority of each server, and the adjustment factor matched with the priority of each server at the prediction time. Based on the adjusted power of each server, the maximum power used by each server in the time period from the prediction time to the next prediction time is determined. This maximum power usage limits the power of the server to be less than the maximum power usage of the server in the time period from the arrival of the prediction time to the next prediction time, so as to ensure the power safety of the target rack in the time period from the prediction time to the next prediction time. Based on this, by dynamically adjusting the maximum power usage of each server on the target rack in the future time period, this embodiment can flexibly plan the number of servers to be racked according to the rack space of the target rack, which effectively avoids the problem of rack space waste, thereby achieving both full utilization of rack space and ensuring rack power safety.
[0028] Furthermore, in this embodiment, the predicted power of the server is dynamically adjusted by introducing an adjustment factor and the server priority. The adjustment factor indicates the power adjustment range corresponding to the matched priority. At the same prediction time, the power adjustment range corresponding to the lower priority is greater than the power adjustment range corresponding to the higher priority. Moreover, the larger the adjustment factor, the greater the difference between the power adjustment range corresponding to the lower priority and the power adjustment range corresponding to the higher priority. In this way, by suppressing the maximum power used by the lower priority server, the operating performance of the higher priority server can be guaranteed to the greatest extent.
[0029] The following is a further description of how to adjust the prediction power of the server at the prediction time in step 101 above: As an example, see Figure 2As shown, the above-mentioned adjustment of the server's predicted power at the prediction time is based on the rated power, the predicted power of each server at the prediction time, the server's priority, and the adjustment factor matching the server's priority at the prediction time, to obtain the adjusted power. In specific implementation, this may include the following steps: Step 201: Based on the server's priority and the adjustment factor that matches the server's priority at the prediction time, determine the server's power adjustment weight; the server's power adjustment weight is negatively correlated with the server's priority and positively correlated with the adjustment factor corresponding to the server.
[0030] In this embodiment, as an example, the power adjustment weight of the server can be determined according to the following formula: ;in, This indicates the power adjustment weight of the server; This indicates the adjustment factor that matches the priority of the server at the predicted time. This indicates the priority of the server.
[0031] Optionally, in this embodiment, the priority of each server can be flexibly set based on actual application needs, and is not specifically limited here. For example, it can be set based on the business that the server is running. Servers running critical business can be given a higher priority, while servers running non-critical business can be given a lower priority. The priority of any server can be represented by a value in the range [0, 1]. The larger the value, the higher the priority, and the lower the value, the lower the priority.
[0032] In this embodiment, there are many ways to determine the adjustment factor that matches the priority of the server at the predicted time, and there is no specific limitation here.
[0033] For example, as an implementation, an adjustment factor can be pre-set for different time periods within a day (i.e., 24 hours) based on historical experience and actual application needs. For instance, during the period from 8 a.m. to 6 p.m., other departments besides the server room, such as business departments, also need electricity, and the server room's power supply is tight at this time. Therefore, a larger adjustment factor can be set for the period from 8 a.m. to 6 p.m. during the other time periods of the day, other departments besides the server room usually do not need electricity, and the server room's power supply is not tight at this time. Therefore, a smaller adjustment factor can be set for these other time periods.
[0034] Based on this, this embodiment can maintain a time period-adjustment factor mapping relationship, so that when it is necessary to dynamically determine the adjustment factor, the adjustment factor corresponding to the time period that matches the prediction time in the recorded time period-adjustment factor mapping relationship is used as the adjustment factor that matches the priority of the server at the prediction time.
[0035] For example, as an embodiment, an adjustment factor matching the priority of a server at the prediction time can be determined based on the difference between the sum of the predicted power of each server at the prediction time and the rated power; wherein the adjustment factor matching the priority of the server at the prediction time is positively correlated with the difference.
[0036] Optionally, in this embodiment, the adjustment factor that matches the priority of the server at the prediction time based on the difference between the sum of the predicted power of each server at the prediction time and the rated power of the target rack can be implemented as follows: after obtaining the difference between the sum of the predicted power of each server at the prediction time and the rated power of the target rack, if the ratio of the difference to the rated power of the target rack is within a first range (e.g., ... If the first setting value, such as 1 or 2, is used as an adjustment factor to match the priority of the server at the predicted time, then the ratio of the difference to the rated power of the target rack is within the second range (e.g., ...). If the second setting value, such as 3 or 4, is used as an adjustment factor to match the priority of the server at that prediction time, then the ratio of the difference to the rated power of the target rack is within the third range (e.g., ...). If the third setting value (e.g., 5 or 6) is used as an adjustment factor to match the priority of the server at that prediction time, then the third setting value can be used as an adjustment factor.
[0037] Step 202: Based on the rated power of the target rack, the predicted power of each server at the prediction time, and the power adjustment weight of the server, adjust the predicted power of the server at the prediction time to obtain the adjusted power.
[0038] In this embodiment, as an example, the predicted power of the server at the prediction time can be adjusted according to the following formula to obtain the adjusted power: ;in, This indicates the adjusted power. This indicates the prediction power of the server at that prediction time; This indicates the power adjustment weight of the server; Indicates the rated power of the target rack; This represents the sum of the prediction power of each server at that prediction time.
[0039] The following is a further description of step 102 above: In this embodiment, as one example, the above-mentioned determination of the maximum power used by each server during the time period from the current prediction time to the next prediction time based on the adjusted power of each server can be implemented in a specific way, for example, by directly using the adjusted power of each server as the maximum power used by each server during the time period from the current prediction time to the next prediction time.
[0040] Accordingly, in this embodiment, the above-mentioned determination of the maximum power used by each server in the time period from the prediction time to the next prediction time based on the prediction power of each server at the prediction time can be implemented in a specific way, for example, by directly using the prediction power of each server at the prediction time as the maximum power used by each server in the time period from the prediction time to the next prediction time.
[0041] The following describes how to predict the power of each server on the target rack at each prediction time within a future set time period: In this embodiment, as an example, the power of any server on the target rack is periodically collected; for example, the power of the server can be collected every 30 seconds or 1 minute.
[0042] Subsequently, for the power of the server collected at each collection moment, if it is determined that the power of the server collected at that collection moment meets the preset abnormal data conditions based on the predicted power of the server at that collection moment, it is determined whether the power of the server collected at that collection moment meets the preset abnormal correction conditions based on the CPU utilization and memory usage of the server at that collection moment. If so, the power of the server collected at that collection moment is adjusted and recorded based on the power of the server collected at the previous N historical collection moments, where N is greater than 1.
[0043] Optionally, in this embodiment, the aforementioned preset abnormal data conditions may be, for example: ; in, This indicates the power of the server collected at that particular moment. This indicates the predicted power of the server at that acquisition time. This threshold is dynamically adjusted based on the server's historical power fluctuations, effectively preventing false alarms due to abnormal data. For example, if the standard deviation of the server's historical power collected within a period prior to the current data collection time (e.g., 5 minutes prior to the current data collection time) is greater than or equal to the set standard deviation threshold, then a set value, such as 5% or 10%, is added to the preset base threshold (e.g., 10% or 15%) to obtain... If the standard deviation of the server's historical power collected within a period prior to the collection time is greater than or less than a set standard deviation threshold, then the preset baseline threshold will be used as the baseline threshold. .
[0044] Optionally, in this embodiment, the aforementioned preset anomaly correction conditions may be, for example, that the absolute value of the difference between the CPU utilization of the server at the current acquisition time and the CPU utilization of the server at the previous acquisition time is less than a set CPU threshold (such as 5% or 10%), and the absolute value of the difference between the memory occupancy of the server at the current acquisition time and the memory occupancy of the server at the previous acquisition time is less than a set memory threshold (such as 5% or 10%).
[0045] Optionally, in this embodiment, adjusting the server power collected at the current collection time based on the server power collected at N historical collection times prior to the current collection time can be implemented in a specific way, for example, as follows: First, obtain the power of the server collected at the current collection time and at N historical collection times prior to the current collection time, and then calculate the standard deviation of the obtained power and the average value of the obtained power.
[0046] Then, based on the calculated standard deviation and average value, the power offset value corresponding to the power of the server collected at that time is determined, such as: the power offset value = (the power of the server collected at that time - the above average value) / the above standard deviation.
[0047] If the absolute value of the above power offset is greater than or equal to the set offset threshold (such as 2 or 3), then it is determined that the power of the server collected at this acquisition time needs to be smoothed. In this case, the median of the power of the server collected at this acquisition time and at the N historical acquisition times before this acquisition time can be used as the adjusted power of the server collected at this acquisition time.
[0048] If the absolute value of the above power offset is less than the set offset threshold, the power of the server collected at that time will remain unchanged.
[0049] In this embodiment, as an example, see [example]. Figure 3As shown, the predicted power of each server at this prediction time can be determined through the following steps: Step 301: For each server, determine the historical power change trend of the server based on the historical power collected at each historical collection time within the historical set time period; the historical power change trend of the server refers to the trend of the historical power of the server collected within the historical set time period changing periodically over time.
[0050] In this embodiment, during normal server operation, the services it runs exhibit a certain periodic variation pattern, thus the server's power also shows a periodic variation pattern over time. Based on this, as an example, a Key Performance Indicator (KPI) algorithm can be used to analyze the periodic variation trend of the server's historical power collected at various historical data collection points within a set historical time period. This analysis yields the server's historical power variation trend, which could be a curve showing the periodic variation of the server's historical power over time within a set historical time period. The set historical time period can be flexibly set based on actual application requirements.
[0051] The KPI algorithm is a time-series prediction algorithm that can be used to predict data for a future period based on historical data. This embodiment does not specifically limit how the KPI algorithm is used to predict the historical power change trend of the server based on the collected historical power data.
[0052] Step 302: Based on the historical power change trend of the server, determine the predicted power of the server at the prediction time.
[0053] In this embodiment, as one example, the above-mentioned determination of the predicted power of the server at the prediction time based on the historical power change trend of the server can be implemented as follows: assuming that the historical power change trend of the server is a curve of the historical power of the server changing periodically over a set historical time period, then a matching time that matches the prediction time can be obtained from each period of the curve. The difference between the matching time obtained in any period and the prediction time is an integer multiple of the period size. Then, based on the historical power corresponding to at least one matching time, the predicted power of the server at the prediction time is determined. For example, the average of the historical power corresponding to at least one matching time can be used as the predicted power of the server at the prediction time, etc., and there is no specific limitation here.
[0054] To facilitate understanding of the specific implementation process of the above server power control method, specific embodiments are described below.
[0055] This embodiment provides a server power control method that dynamically adjusts the maximum power used by each server on the target rack in a future time period to achieve an optimal balance between rack space utilization and power safety. In other words, it can maximize rack space utilization while ensuring rack power safety.
[0056] In this embodiment, the method can be applied to centralized server management software deployed in the operation and maintenance server. The centralized server management software can support the relationship management between racks and servers. For example, users can add servers to be racked at the corresponding position of the rack through the software, and can set the rated power of the rack, etc., which are not specifically limited here.
[0057] As an example, the specific implementation process of this method is as follows: The centralized server management software can periodically (e.g., every 30 seconds) collect and store the power, CPU utilization, and memory usage of each server by calling the Baseboard Management Controller (BMC) interface provided by each server on the target rack.
[0058] In this embodiment, for the power of each server collected at each collection moment, if it is determined that the power of the server collected at that collection moment meets the preset abnormal data conditions based on the predicted power of the server at that collection moment, then based on the CPU utilization and memory usage of the server at that collection moment, it is determined whether the power of the server collected at that collection moment meets the preset abnormal correction conditions. If yes, then based on the power of the server collected at the previous N historical collection moments, the power of the server collected at that collection moment is adjusted and recorded. If no, then no adjustment is needed, and it is recorded directly.
[0059] Based on the above description, for each server, the centralized server management software can periodically (e.g., every 30 minutes) use a KPI algorithm to analyze the periodic change trend of the historical power collected at each historical acquisition time within a historically set time period, so as to obtain the historical power change trend of the server. Based on the historical power change trend of the server, the predicted power of the server at each prediction time within a future set time period (e.g., the next 30 minutes) can be determined.
[0060] Among them, at least one prediction time within a set future time period can be determined based on the power acquisition frequency, such as determining a prediction time every 30 seconds.
[0061] Based on this, for each predicted moment within the aforementioned future time period, if the sum of the predicted powers of all servers at that predicted moment is found to be greater than the rated power configured for the target rack, then: for each server, based on the rated power, the predicted power of each server at that predicted moment, the server's priority, and the adjustment factor matching the server's priority at that predicted moment, the predicted power of that server at that predicted moment is adjusted to obtain the adjusted power; and based on the adjusted power of each server, the maximum power used by each server during the time period from that predicted moment to the next predicted moment is determined, wherein the maximum power used by any server limits the power of that server to be less than its maximum power used during the time period from the arrival of that predicted moment to the next predicted moment.
[0062] If the sum of the predicted power of each server on the target rack at the predicted time is less than or equal to the rated power configured for the target rack, the maximum power used by each server during the time period from the predicted time to the next predicted time can be determined directly based on the predicted power of each server at the predicted time.
[0063] This concludes the description of the method provided in the embodiments of this application. The apparatus provided in the embodiments of this application will now be described: As an example, this embodiment also provides a server power control device. For example, see... Figure 4 , Figure 4 This is a schematic diagram of a server power control device provided in an embodiment of this application. The device corresponds to… Figure 1 The method flow is shown below. Figure 5 As shown, the server power control device 400 is applied to electronic devices, and the server power control device 400 includes: Control module 401 is configured to, for each prediction time, if the sum of the predicted powers of all servers on the target rack at that prediction time is found to be greater than the rated power configured for the target rack, then: For each server, based on the rated power, the predicted power of each server at the prediction time, the server's priority, and the adjustment factor matched with the server's priority at the prediction time, the predicted power of the server at the prediction time is adjusted to obtain the adjusted power; the adjustment factor indicates the power adjustment range corresponding to the matched priority; at the same prediction time, the power adjustment range corresponding to the lower priority is greater than the power adjustment range corresponding to the higher priority, and the larger the adjustment factor, the greater the difference between the power adjustment range corresponding to the lower priority and the power adjustment range corresponding to the higher priority. Based on the adjusted power of each server, the maximum power used by each server during the period from the current prediction time to the next prediction time is determined. The maximum power used by any server is limited to the fact that the power of that server is less than its maximum power used during the period from the arrival of the current prediction time to the next prediction time.
[0064] As one embodiment, the step of adjusting the predicted power of a server at the prediction time based on the rated power, the predicted power of each server at the prediction time, the priority of the server, and the adjustment factor matching the priority of the server at the prediction time, to obtain the adjusted power includes: Based on the server's priority and the adjustment factor that matches the server's priority at the prediction time, the server's power adjustment weight is determined; the server's power adjustment weight is negatively correlated with the server's priority, and positively correlated with the adjustment factor corresponding to the server. Based on the rated power, the predicted power of each server at the prediction time, and the power adjustment weight of the server, the predicted power of the server at the prediction time is adjusted to obtain the adjusted power.
[0065] As an example, the adjustment factor matching the server's priority at this prediction time is determined through the following steps: The adjustment factor corresponding to the time period that matches the prediction time in the recorded time period-adjustment factor mapping relationship is used as the adjustment factor that matches the priority of the server at the prediction time. or, Based on the difference between the sum of the predicted power of each server at the prediction time and the rated power, an adjustment factor matching the priority of the server at the prediction time is determined; the adjustment factor matching the priority of the server at the prediction time is positively correlated with the difference.
[0066] As one embodiment, the control module 401 is further configured to: If the sum of the predicted power of each server on the target rack at the predicted time is found to be greater than the rated power, then based on the predicted power of each server at the predicted time, the maximum power used by each server during the time period from the predicted time to the next predicted time is determined.
[0067] As one embodiment, the device further includes: The data acquisition module is used to periodically collect the power of any server on the target rack. The adjustment module is used to adjust the power of the server collected at each acquisition time. When it is determined that the power of the server collected at the acquisition time meets the preset abnormal data conditions based on the predicted power of the server at the acquisition time, it is determined whether the power of the server collected at the acquisition time meets the preset abnormal correction conditions based on the CPU utilization and memory usage of the server at the acquisition time. If so, the power of the server collected at the acquisition time is adjusted based on the power of the server collected at N historical acquisition times before the acquisition time and recorded, where N is greater than 1.
[0068] As an example, the predicted power of each server at that prediction time is determined by the following steps: For each server, the historical power change trend of the server is determined based on the historical power collected at each historical collection time within the historical set time period; the historical power change trend of the server refers to the trend of the historical power of the server changing periodically over time within the historical set time period. Based on the server's historical power change trend, determine the predicted power of the server at the predicted time.
[0069] This concludes the process. Figure 4 Structural description of the device shown.
[0070] The specific implementation process of the functions and roles of each module in the above device can be found in the implementation process of the corresponding steps in the above method, and will not be repeated here.
[0071] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this application according to actual needs. Those skilled in the art can understand and implement this without creative effort.
[0072] Please see Figure 5 This is a schematic diagram of the hardware structure of an electronic device provided as an exemplary embodiment of this application. The electronic device includes a processor and a computer-readable storage medium; the computer-readable storage medium stores a plurality of computer program instructions, which, when executed by the processor, implement the method disclosed in the above example of this application. Depending on the actual function of the electronic device, other hardware may also be included, which will not be elaborated further.
[0073] Based on the same concept as the above method, this application also provides a computer-readable storage medium storing a plurality of computer program instructions, which, when executed by a processor, can implement the method disclosed in the above examples of this application.
[0074] For example, the aforementioned computer-readable storage medium can be any electronic, magnetic, optical, or other physical storage device that can contain or store information such as executable instructions, messages, etc. For instance, computer-readable storage media can be: RAM (Random Access Memory), volatile memory, non-volatile memory, flash memory, storage drives (such as hard disk drives), solid-state drives, any type of storage disk (such as optical discs, DVDs, etc.), or similar storage media, or combinations thereof.
[0075] The above are merely preferred embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.
Claims
1. A method of server power control, the method comprising: The method comprises: For each prediction time, if it is found that the sum of the predicted powers of the servers on the target rack at the prediction time is greater than the rated power configured for the target rack, then: For each server, based on the rated power, the predicted power of each server at the prediction time, the priority of the server, and the adjustment factor matched with the priority of the server at the prediction time, the predicted power of the server at the prediction time is adjusted to obtain an adjusted power; the adjustment factor indicates the power adjustment amplitude corresponding to the matched priority; at the same prediction time, the power adjustment amplitude corresponding to a low priority is greater than the power adjustment amplitude corresponding to a high priority, and the greater the adjustment factor, the greater the difference between the power adjustment amplitude corresponding to a low priority and the power adjustment amplitude corresponding to a high priority. Based on the adjusted powers of the servers, the maximum use powers of the servers in a time period from the prediction time to a next prediction time are determined; the maximum use power of any server limits the power of the server in the time period from the prediction time to the next prediction time.
2. The method of claim 1, wherein, The adjustment of the predicted power of the server at the prediction time based on the rated power, the predicted power of each server at the prediction time, the priority of the server, and the adjustment factor matched with the priority of the server at the prediction time to obtain an adjusted power comprises: Based on the priority of the server and the adjustment factor matched with the priority of the server at the prediction time, a power adjustment weight of the server is determined; the power adjustment weight of the server is negatively correlated with the priority of the server, and the power adjustment weight of the server is positively correlated with the adjustment factor corresponding to the server. The adjustment of the predicted power of the server at the prediction time based on the rated power, the predicted power of each server at the prediction time, and the power adjustment weight of the server to obtain an adjusted power.
3. The method according to claim 1 or 2, characterized in that, The adjustment factor matched with the priority of the server at the prediction time is determined by the following steps: The adjustment factor corresponding to the time period matched with the prediction time in the recorded time period-adjustment factor mapping relationship is taken as the adjustment factor matched with the priority of the server at the prediction time. Or, Based on the difference between the sum of the predicted powers of the servers at the prediction time and the rated power, the adjustment factor matched with the priority of the server at the prediction time is determined; the adjustment factor matched with the priority of the server at the prediction time is positively correlated with the difference.
4. The method of claim 1, wherein, The method further comprises: If it is found that the sum of the predicted powers of the servers on the target rack at the prediction time is greater than the rated power, then based on the predicted powers of the servers at the prediction time, the maximum use powers of the servers in a time period from the prediction time to a next prediction time are determined.
5. The method of claim 1, wherein, The method further comprises: Periodically collecting the power of any server on the target rack. For the power of the server collected at each collection time, when it is determined that the power of the server collected at the collection time satisfies a preset abnormal data condition based on the obtained predicted power of the server at the collection time, it is determined whether the power of the server collected at the collection time satisfies a preset abnormal correction condition based on the central processing unit (CPU) utilization and the memory occupation of the server at the collection time, and if so, the power of the server collected at the collection time is adjusted based on the powers of the server collected at N historical collection times before the collection time, and the adjusted power is recorded, where N is greater than 1.
6. The method of claim 5, wherein, The predicted power of each server at the prediction time is determined by the following steps: For each server, a historical power change trend of the server is determined according to historical powers of the server collected at historical collection times within a historical setting time period; the historical power change trend of the server refers to a trend of periodic change of the historical powers of the server collected within the historical setting time period with time; The predicted power of the server at the prediction time is determined according to the historical power change trend of the server.
7. A server power control apparatus, characterized by comprising: The device comprises: The control module is configured to, for each prediction time, if it is found that the sum of the predicted powers of the servers on the target rack at the prediction time is greater than the rated power configured for the target rack, then: For each server, the predicted power of the server at the prediction time is adjusted based on the rated power, the predicted power of each server at the prediction time, the priority of the server, and an adjustment factor matched with the priority of the server at the prediction time, to obtain an adjusted power; the adjustment factor indicates a power adjustment range corresponding to the matched priority; at the same prediction time, the power adjustment range corresponding to a low priority is greater than the power adjustment range corresponding to a high priority, and the greater the adjustment factor, the greater the difference between the power adjustment range corresponding to the low priority and the power adjustment range corresponding to the high priority. Based on the adjusted powers of the servers, the maximum usage power of each server within a time period from the prediction time to a next prediction time is determined, and the maximum usage power of any server limits the power of the server within the time period from the prediction time to the next prediction time.
8. The apparatus of claim 7, wherein, The adjustment of the predicted power of the server at the prediction time based on the rated power, the predicted power of each server at the prediction time, the priority of the server, and the adjustment factor matched with the priority of the server at the prediction time to obtain the adjusted power comprises: Based on the priority of the server and the adjustment factor matched with the priority of the server at the prediction time, a power adjustment weight of the server is determined; the power adjustment weight of the server is negatively correlated with the priority of the server, and the power adjustment weight of the server is positively correlated with the adjustment factor corresponding to the server. adjusting the predicted power of the server at the predicted time based on the rated power, the predicted power of each server at the predicted time, and the power adjustment weight of the server, to obtain an adjusted power; and / or, the adjustment factor matched with the priority of the server at the predicted time is determined by the following steps: taking the adjustment factor corresponding to the time period matched with the predicted time in the recorded time period-adjustment factor mapping relationship as the adjustment factor matched with the priority of the server at the predicted time; or, determining the adjustment factor matched with the priority of the server at the predicted time based on a difference between the sum of the predicted power of each server at the predicted time and the rated power; the adjustment factor matched with the priority of the server at the predicted time is positively correlated with the difference; and / or, the control module is further configured to: if it is found that the sum of the predicted power of each server on the target rack at the predicted time is greater than the rated power, determining the maximum usage power of each server in the time period from the predicted time to the next predicted time based on the predicted power of each server at the predicted time; and / or, the device further comprises: a collection module configured to periodically collect the power of any server on the target rack; an adjustment module configured to, for the power of the server collected at each collection time, determine whether the power of the server collected at the collection time meets a preset abnormal data condition based on the predicted power of the server at the collection time obtained, and if so, determine whether the power of the server collected at the collection time meets a preset abnormal correction condition based on the central processing unit (CPU) utilization and memory occupancy of the server at the collection time, and if so, adjust the power of the server collected at the collection time based on the power of the server collected at N historical collection times before the collection time, and record the adjusted power, wherein N is greater than 1; and / or, the predicted power of each server at the predicted time is determined by the following steps: for each server, determining a historical power change trend of the server based on the historical power of the server collected at each historical collection time in a historical set time period; the historical power change trend of the server refers to a trend of the historical power of the server collected in the historical set time period changing periodically with time; determining the predicted power of the server at the predicted time based on the historical power change trend of the server.
9. An electronic device, comprising: The electronic device comprises: a processor; and a computer readable storage medium having computer program instructions stored therein, the computer program instructions, when executed by the processor, causing the processor to perform the steps in any one of the methods of claims 1 to 6.
10. A computer-readable storage medium, characterized in that, The computer readable storage medium has computer program instructions stored thereon, the computer program instructions, when executed by a processor, causing the processor to perform the steps in any one of the methods of claims 1 to 6.