Method, system, medium and program product for balancing host power consumption and temperature

Abstract

A method, system, medium and program product for balanced control of host power consumption and temperature, relating to the field of electrical digital data processing, in which a plurality of heat dissipation ducts are constructed according to air flow data; temperature data in each heat dissipation duct is collected, and the temperature response rate and temperature decay rate of each heat dissipation duct are calculated respectively; computing tasks are divided into first-class computing tasks and second-class computing tasks based on the temperature response rate; migration rules for the first-class computing tasks and the second-class computing tasks are established according to the temperature decay rate; when it is detected that the temperature in any heat dissipation duct is greater than the upper temperature threshold, the first-class computing tasks being executed in the heat dissipation duct are migrated to the target heat dissipation duct whose temperature is lower than the preset temperature threshold; the second-class computing tasks being executed in the target heat dissipation duct are migrated to the heat dissipation duct. This application improves the accuracy of controlling the transfer and accumulation of heat inside the device, thereby improving the accuracy of temperature control management.

Description

Technical Field

[0001] The present application belongs to the field of electronic digital data processing, and in particular, relates to a method, system, medium and program product for balancing host power consumption and temperature. Background Art

[0002] With the rapid development of information technology, the processing performance of host devices is constantly improving, while power consumption management is also facing severe challenges. High power consumption will not only cause the device temperature to rise, but also increase the burden on the cooling system. Excessive temperature may cause system instability or device damage. Therefore, how to achieve balanced control of temperature and power consumption while ensuring host performance has become an important issue that needs to be solved urgently.

[0003] In related technologies, the temperature and power consumption status of each core component of the host can be monitored in real time to establish a correlation mapping between temperature and power consumption, and dynamically adjust the operating parameters of the processor based on the mapping relationship. The system can adaptively adjust the power consumption distribution while maintaining stable performance, balancing the temperature and performance requirements of the device.

[0004] However, since the transfer of heat inside the host has a significant cumulative effect and hysteresis, a single control strategy is difficult to accurately grasp the thermal coupling between components, which can easily cause problems such as local overheating or excessive frequency reduction. Summary of the invention

[0005] The present application provides a method, system, medium and program product for balanced control of host power consumption and temperature, which are used to improve the accuracy of controlling the transfer and accumulation of heat inside the device, thereby improving the accuracy of temperature control management.

[0006] In a first aspect, the present application provides a method for balancing host power consumption and temperature, wherein a plurality of heat dissipation ducts are constructed inside the host according to air flow data inside the host collected in real time by a temperature sensor, and each heat dissipation duct passes through at least two functional components, and the functional components include a processor, a memory, and a chipset;

[0007] Collect temperature data in each heat dissipation duct, and calculate the temperature response rate and temperature decay rate of each heat dissipation duct respectively. The temperature response rate represents how fast the temperature in the heat dissipation duct rises, and the temperature decay rate represents how fast the temperature in the heat dissipation duct drops;

[0008] Based on the temperature response rate, the computing tasks of the host are divided into a first type of computing tasks and a second type of computing tasks, wherein the first type of computing tasks causes the processor temperature to rise at a rate greater than a preset temperature threshold during execution, and the second type of computing tasks causes the processor temperature to rise at a rate less than the preset temperature threshold during execution;

[0009] According to the temperature decay rate of each heat dissipation air duct, a migration rule for the first type of computing tasks and the second type of computing tasks is established, and the migration rule is used to determine the transfer timing and transfer path of the computing tasks between different heat dissipation air ducts;

[0010] When it is detected that the temperature in any heat dissipation duct is greater than the upper temperature threshold, the first type of computing task being executed in the heat dissipation duct is migrated to the target heat dissipation duct whose temperature is lower than the preset temperature threshold; the second type of computing task being executed in the target heat dissipation duct is migrated to the heat dissipation duct.

[0011] By adopting the above technical solution, by building multiple heat dissipation ducts inside the host and obtaining the temperature response rate and temperature attenuation rate, the heat dissipation characteristics of different areas can be accurately grasped. By classifying computing tasks based on the temperature response rate, it is possible to identify the types of tasks that have a greater and lesser impact on temperature. Task migration rules are formulated in combination with the temperature attenuation rate. When a temperature anomaly is detected, the high-temperature generation task is migrated to the low-temperature area, and the low-temperature generation task in the low-temperature area is replaced to the original area, forming a virtuous cycle of temperature balance, improving the accuracy of controlling the transfer and accumulation of heat inside the device, and thereby improving the accuracy of temperature control management.

[0012] In combination with some embodiments of the first aspect, in some embodiments, a plurality of heat dissipation ducts are constructed inside the host according to the air flow data inside the host collected in real time by the temperature sensor, specifically including:

[0013] Determine a temperature data sequence based on the air flow data inside the host collected in real time by the temperature sensor;

[0014] Construct a temperature field distribution model inside the host based on the temperature data sequence;

[0015] Based on the temperature field distribution model, the minimum spanning tree algorithm is used to determine the optimal connection path between each functional component;

[0016] Functional components with the same optimal connection path are combined into one heat dissipation duct to obtain a plurality of heat dissipation ducts.

[0017] By adopting the above technical solution, a temperature field distribution model is constructed using a temperature data sequence, and the optimal connection path between functional components is determined based on the minimum spanning tree algorithm, so that the constructed heat dissipation duct has the optimal heat transfer efficiency. The temperature field distribution model reflects the actual temperature distribution inside the host. The minimum spanning tree algorithm ensures that the connection path between functional components is the shortest, reducing the loss of heat during the transfer process. Functional components with the same optimal connection path are combined into a heat dissipation duct, so that the temperature transfer in each duct is smoother, the heat dissipation efficiency is improved, and the heat dissipation system's adaptability to temperature changes and the heat dissipation effect are improved.

[0018] In conjunction with some embodiments of the first aspect, in some embodiments, the computing tasks of the host are divided into a first type of computing tasks and a second type of computing tasks based on the temperature response rate, specifically including:

[0019] Construct resource consumption feature vectors of computing tasks, which include processor occupancy, memory occupancy, and input and output frequency;

[0020] Adopt adaptive clustering algorithm to cluster the feature vectors and obtain several cluster centers;

[0021] Calculate the temperature response characteristic curve corresponding to each type of cluster center;

[0022] Extract the key inflection points of the temperature response characteristic curve, and determine the critical interval of temperature rise based on the key inflection points;

[0023] The computing tasks are divided into the first category computing tasks and the second category computing tasks according to the critical interval.

[0024] By adopting the above technical solution, by constructing the resource consumption feature vector of the computing task and performing cluster analysis, the system can deeply understand the characteristics of different tasks from the resource occupation dimension. The temperature response characteristic curve of the cluster center is analyzed and the key inflection points are extracted to accurately identify the key range of temperature change during task execution, and a direct correlation is established between the resource consumption characteristics of the task and its temperature impact characteristics, reducing the deviation that may be caused by classification based on a single feature. By extracting the key inflection points of the temperature response characteristic curve to determine the critical range of temperature rise, the system can more accurately predict the degree of impact of task execution on temperature, thereby achieving more accurate classification of computing tasks.

[0025] In combination with some embodiments of the first aspect, in some embodiments, after migrating the second type of computing tasks being executed in the target heat dissipation duct to the heat dissipation duct, the method further includes:

[0026] Obtaining the performance parameters of each functional component during the migration of the computing task, and determining the fluctuation time interval and fluctuation amplitude of the performance parameters;

[0027] Combining the first type of computing tasks and the second type of computing tasks into a plurality of computing task units according to the performance parameters, wherein the plurality of computing tasks contained in each computing task unit respectively occupy different types of computing resources;

[0028] The migration execution timing of each computing task unit is set based on the fluctuation time interval, and the migration execution timing makes the migration time interval of adjacent computing task units greater than a preset time threshold;

[0029] In the process of executing the migration execution sequence, computing task units whose performance fluctuation amplitude is greater than the average value of the fluctuation amplitude and computing task units whose performance fluctuation amplitude is less than the average value of the fluctuation amplitude are alternately migrated;

[0030] When the fluctuation amplitude of the performance parameter exceeds the preset time threshold, the migration time of the subsequent computing task unit is postponed.

[0031] By adopting the above technical solution, the system achieves balanced utilization of computing resources by obtaining the performance parameter fluctuation characteristics during task migration and combining different types of computing tasks into computing task units. The migration execution timing set based on the performance fluctuation characteristics ensures the smooth progress of the task migration process. The strategy of alternately migrating computing task units with different performance fluctuation amplitudes smoothes the overall performance fluctuations. When performance fluctuations beyond expectations occur, by postponing the migration time of subsequent tasks, a buffer time for performance recovery is reserved for the system, maintaining the stability of system performance while achieving temperature balance, and improving the stability of the system during temperature control.

[0032] In combination with some embodiments of the first aspect, in some embodiments, the migration execution sequence of each computing task unit is set based on the fluctuation time interval, specifically including:

[0033] Obtain the performance recovery time of each computing task unit during the historical migration process. The performance recovery time is the time required for the performance parameters to recover to the preset benchmark level after the computing task migration is completed.

[0034] The computing task units are divided into several migration batches according to the performance recovery time, and the sum of the performance recovery time of the computing task units in each migration batch is less than the fluctuation time interval;

[0035] Determine the execution order of each migration batch so that the migration time interval between adjacent migration batches is not less than the maximum performance recovery time in the migration batch;

[0036] In each migration batch, migration time points are assigned to computing task units in the order of performance fluctuation from small to large;

[0037] When the actual performance recovery time of any computing task unit is greater than the historical performance recovery time, all migration time points after the migration batch will be postponed;

[0038] Generate a migration execution sequence including the migration time points of each computing task unit.

[0039] By adopting the above technical solution and obtaining the historical performance recovery time of the computing task unit, the system can accurately evaluate the duration of the impact of each task migration on the system performance. The task units are divided into multiple migration batches based on the performance recovery time to ensure that the total performance recovery time of the tasks in each batch does not exceed the fluctuation time interval. By setting the time interval between adjacent batches to be no less than the maximum performance recovery time within the batch, the system can maintain relative stability of performance during task migration. Migration time points are allocated within the batch from small to large according to the performance fluctuation amplitude, so that tasks with large performance fluctuations can obtain sufficient recovery time. When the actual recovery time exceeds expectations, the migration rhythm is dynamically adjusted by postponing subsequent time points, which ensures the adaptability of the system and improves the stability and reliability of the system operation.

[0040] In combination with some embodiments of the first aspect, in some embodiments, after generating the migration execution sequence including the migration time points of each computing task unit, the method further includes:

[0041] Calculate the migration duration of each computing task unit under the migration execution sequence, where the migration duration is the time required for the computing task to start migration and complete migration;

[0042] Determine the migration priority of each computing task unit according to the migration duration and performance recovery time;

[0043] When it is detected that the temperature change rate of the heat dissipation air duct exceeds a preset temperature threshold, the migration order of the computing task units having the same migration time point is adjusted.

[0044] By adopting the above technical solution, the migration priority of the task unit is determined by calculating the migration duration and performance recovery time, and the system can quantitatively evaluate the impact of each task migration on the overall performance. When it is detected that the temperature change rate of the heat dissipation duct exceeds the threshold, the migration order of the tasks at the same time point is adjusted in a targeted manner, so that the system can perform local optimization on the basis of maintaining the original migration plan framework, and can flexibly adjust the task execution order when facing temperature anomalies, and optimize the temperature control effect without affecting the overall migration plan. Through the comprehensive consideration of the two key indicators of migration duration and performance recovery time, the system can achieve a dynamic balance between performance and temperature during the task migration process, improve the accuracy and timeliness of temperature control, and ensure a smooth transition of system performance.

[0045] In conjunction with some embodiments of the first aspect, in some embodiments, adjusting the migration order of computing task units having the same migration time point specifically includes:

[0046] Obtain the resource occupancy type and occupancy ratio of each computing task unit;

[0047] Calculate the influence coefficient of each computing task unit on the temperature of the heat dissipation duct;

[0048] Sort the computing task units with the same migration time point according to the influence coefficient;

[0049] Arrange the computing task units whose influence coefficient is greater than a preset influence coefficient threshold to be executed in a time period when the temperature is lower than the target temperature;

[0050] When the temperature change rate drops below a preset temperature threshold, the migration sequence is restored.

[0051] By adopting the above technical solution, by analyzing and calculating the resource occupancy type and occupancy ratio of the task unit, and combining the sorting of the influence coefficient of the heat dissipation duct temperature, the system can accurately identify and quantify the contribution of different tasks to temperature changes. Scheduling tasks with high influence coefficients to be executed during time periods with lower temperatures can prevent drastic temperature fluctuations. When the temperature change rate drops below the threshold, the original migration order is restored to ensure the effectiveness of temperature control, enabling the system to achieve more accurate temperature control during task migration, reducing the impact of drastic temperature fluctuations on system stability, and improving the system's operating efficiency.

[0052] In a second aspect, an embodiment of the present application provides a balanced control system for host power consumption and temperature, which includes: one or more processors and a memory; the memory is coupled to one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and one or more processors call the computer instructions to enable the system to execute the method described in the first aspect and any possible implementation method of the first aspect.

[0053] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, comprising instructions, which, when executed on a system, causes the system to execute the method described in the first aspect and any possible implementation of the first aspect.

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

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

[0056] 1. The present application provides a method for balanced control of host power consumption and temperature. By constructing multiple heat dissipation ducts inside the host and obtaining the temperature response rate and temperature attenuation rate, the heat dissipation characteristics of different areas can be accurately grasped. By classifying computing tasks based on the temperature response rate, the types of tasks that have a greater and lesser impact on temperature can be identified. Task migration rules are formulated in combination with the temperature attenuation rate. When a temperature anomaly is detected, the high-temperature generation task is migrated to the low-temperature area, and the low-temperature generation task in the low-temperature area is replaced to the original area, forming a virtuous cycle of temperature balance, improving the accuracy of controlling the transfer and accumulation of heat inside the device, and thereby improving the accuracy of temperature control management.

[0057] 2. The present application provides a method for balanced control of host power consumption and temperature. By obtaining the performance parameter fluctuation characteristics during task migration and combining different types of computing tasks into computing task units, the system achieves balanced utilization of computing resources. The migration execution timing set based on the performance fluctuation characteristics ensures the smooth progress of the task migration process. The strategy of alternately migrating computing task units with different performance fluctuation amplitudes smoothes the overall performance fluctuations. When performance fluctuations beyond expectations occur, by postponing the migration time of subsequent tasks, a buffer time for performance recovery is reserved for the system, which maintains the stability of system performance while achieving temperature balance, and improves the stability of the system during temperature control.

[0058] 3. The present application provides a method for balanced control of host power consumption and temperature. By calculating the migration duration and performance recovery time to determine the migration priority of the task unit, the system can quantitatively evaluate the impact of each task migration on the overall performance. When it is detected that the temperature change rate of the heat dissipation duct exceeds the threshold, the migration order of tasks at the same time point is adjusted in a targeted manner, so that the system can perform local optimization on the basis of maintaining the original migration plan framework, and can flexibly adjust the task execution order when facing temperature anomalies, and optimize the temperature control effect without affecting the overall migration plan. Through the comprehensive consideration of the two key indicators of migration duration and performance recovery time, the system can achieve a dynamic balance between performance and temperature during the task migration process, improve the accuracy and timeliness of temperature control, and ensure a smooth transition of system performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a flow chart of a method for balancing host power consumption and temperature in an embodiment of the present application.

[0060] Figure 2 It is a flow chart of a task migration optimization method considering performance fluctuations in an embodiment of the present application.

[0061] Figure 3It is a schematic diagram of the physical device structure of a host power consumption and temperature balance control system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0062] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to be used as limitations to the present application. As used in the specification and appended claims of the present application, the singular expressions "one", "a kind of", "said", "above", "the" and "this" are intended to also include plural expressions, unless there is a clear indication to the contrary in the context. It should also be understood that the term "and / or" used in the present application refers to any or all possible combinations comprising one or more listed items.

[0063] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, unless otherwise specified, "plurality" means two or more.

[0064] The following uses an embodiment and combines Figure 1 , a method for balancing host power consumption and temperature in an embodiment of the present application is described:

[0065] See also Figure 1 , which is a flow chart of a method for balancing host power consumption and temperature in an embodiment of the present application.

[0066] S101, constructing a plurality of heat dissipation ducts inside the host according to the air flow data inside the host collected in real time by the temperature sensor;

[0067] The system constructs a plurality of heat dissipation ducts inside the host according to the air flow data inside the host collected in real time by the temperature sensor, and each heat dissipation duct passes through at least two functional components, and the functional components include a processor, a memory and a chipset. Specifically: a temperature data sequence is determined according to the air flow data inside the host collected in real time by the temperature sensor;

[0068] Construct a temperature field distribution model inside the host based on the temperature data sequence;

[0069] Based on the temperature field distribution model, the minimum spanning tree algorithm is used to determine the optimal connection path between each functional component;

[0070] Functional components with the same optimal connection path are combined into one heat dissipation duct to obtain a plurality of heat dissipation ducts.

[0071] In this step, the system uses the temperature sensor installed inside the host to monitor the air flow inside the host in real time, including parameters such as air temperature and flow rate. Based on this data, the system can analyze the distribution and transfer path of heat inside the host, and then plan several heat dissipation ducts inside the host. Each heat dissipation duct will pass through at least two functional components that generate more heat, such as processors, memory, chipsets, etc., to optimize the heat dissipation path.

[0072] In order to realize the automatic construction of the heat dissipation duct, the system first needs to generate a temperature data sequence based on the data collected by the temperature sensor to reflect the real-time temperature changes in various areas inside the host. Then, the system uses the temperature sequence to construct a three-dimensional temperature field distribution model inside the host, intuitively showing the diffusion of heat inside the host. On this basis, the system can use the minimum spanning tree algorithm, with each functional component as a node and the temperature gradient as a weight, to calculate the optimal heat dissipation path connecting each component. Finally, functional components with similar heat dissipation paths are combined to form several heat dissipation ducts.

[0073] When constructing the cooling duct, if the internal structure of the host is complex and the components are arranged compactly, the layout of the cooling duct may be unreasonable, affecting the cooling effect. To address this problem, the system can verify and optimize the cooling duct through simulation after generating the cooling duct. The system can establish a thermal simulation model inside the host, simulate the cooling performance of the cooling duct under different working conditions, and iteratively adjust the duct layout until the optimal solution is found. In addition, the system can also use artificial intelligence algorithms, such as genetic algorithms, particle swarm algorithms, etc., to intelligently search for the optimal cooling duct layout in massive simulation data to further improve cooling performance.

[0074] S102, collecting temperature data in each heat dissipation duct, and calculating the temperature response rate and temperature decay rate of each heat dissipation duct respectively;

[0075] The system collects temperature data in each heat dissipation duct and calculates the temperature response rate and temperature decay rate of each heat dissipation duct respectively. The temperature response rate represents how fast the temperature in the heat dissipation duct rises, and the temperature decay rate represents how fast the temperature in the heat dissipation duct drops.

[0076] In this step, the system evaluates the performance of the constructed cooling duct. The system collects real-time temperature data in each cooling duct and calculates two key indicators: temperature response rate and temperature decay rate. The temperature response rate reflects how fast the temperature inside the cooling duct rises, and the temperature decay rate reflects how fast the temperature inside the cooling duct drops. By analyzing these two indicators, the system can grasp the heat dissipation performance and thermal response characteristics of each cooling duct, providing a basis for subsequent task scheduling and thermal management.

[0077] In order to obtain accurate temperature response rate and temperature attenuation rate, the system needs to continuously sample the temperature of each heat dissipation duct. Key measurement points of the heat dissipation duct, such as the duct inlet, outlet, and near each functional component, can be selected to arrange a high-sensitivity temperature sensor array. The system collects temperature data at a fixed frequency to form a time-temperature curve. By estimating the slope of the curve in segments, the temperature rise and fall rates can be obtained. The system can also smooth the discrete data points obtained by sampling through segmented fitting methods to reduce the impact of measurement noise and improve the accuracy of rate estimation.

[0078] The temperature distribution inside the cooling duct may be uneven, resulting in large fluctuations in the measured temperature response curve, affecting the accuracy of rate calculation. To solve this problem, the system can adopt a multi-point temperature measurement strategy, arrange sensors at different locations in the cooling duct, and integrate data from multiple measurement points to obtain a more comprehensive and stable temperature characteristic curve. At the same time, the system can introduce algorithms such as Kalman filtering to dynamically track temperature change trends, intelligently eliminate abnormal fluctuations, smooth the temperature curve, and improve the robustness of rate estimation.

[0079] S103, dividing the computing tasks of the host into a first type of computing tasks and a second type of computing tasks based on the temperature response rate;

[0080] The system divides the computing tasks of the host into the first type of computing tasks and the second type of computing tasks based on the temperature response rate. The first type of computing tasks make the processor temperature rise faster than the preset temperature threshold during execution, and the second type of computing tasks make the processor temperature rise faster than the preset temperature threshold during execution. Specifically: construct a resource consumption feature vector of the computing task, which includes processor occupancy, memory occupancy and input and output frequency;

[0081] Adopt adaptive clustering algorithm to cluster the feature vectors and obtain several cluster centers;

[0082] Calculate the temperature response characteristic curve corresponding to each type of cluster center;

[0083] Extract the key inflection points of the temperature response characteristic curve, and determine the critical interval of temperature rise based on the key inflection points;

[0084] The computing tasks are divided into the first category computing tasks and the second category computing tasks according to the critical interval.

[0085] In this step, the system classifies the host's computing tasks according to the temperature response characteristics of the cooling air duct. The system classifies computing tasks that cause the processor temperature to rise rapidly into the first category, and computing tasks that cause the processor temperature to rise slowly into the second category. Through task classification, the system can reasonably schedule different types of tasks according to the real-time heat dissipation capacity of the cooling air duct to prevent the temperature of the overheated air duct from continuing to rise, triggering performance bottlenecks or even safety restrictions.

[0086] In order to realize the automatic classification of computing tasks, the system first needs to extract the resource consumption characteristics of each task and construct a feature vector. The feature vector usually includes indicators such as CPU occupancy, memory occupancy, IO frequency, etc., which reflects the intensity of the task's use of various resources on the host. Then, the system uses adaptive clustering algorithms, such as K-means, DBSCAN, etc., to divide the feature vector space into several clusters, each of which corresponds to a group of tasks with similar resource consumption patterns. Next, the system selects the central tasks of each cluster and executes them in a real or simulated environment to obtain the temperature response characteristic curve caused by them in the heat dissipation duct. By analyzing the key inflection points of the curve, such as the steep rise section and the slow rise section, the system can determine the thermal characteristics of this type of task, whether it belongs to the first type of "high heat" task or the second type of "low heat" task.

[0087] For some complex or new computing tasks, their resource consumption patterns may not be easy to describe with a few feature dimensions, resulting in a decrease in the accuracy of clustering and affecting the reliability of task classification. To address this problem, the system can adopt two strategies: first, using domain knowledge and expert experience to customize personalized feature engineering methods for complex tasks, extract more detailed and comprehensive task attributes, and improve the semantic expression ability of clustering; second, introducing an online learning mechanism to continuously collect its resource consumption data and temperature response data during task execution, and use actual observed data to continuously correct the clustering results in the offline stage, and dynamically optimize the task classification strategy.

[0088] S104, establishing migration rules for the first type of computing tasks and the second type of computing tasks according to the temperature decay rate of each heat dissipation air duct;

[0089] The system establishes migration rules for the first type of computing tasks and the second type of computing tasks according to the temperature decay rate of each cooling air duct. The migration rules are used to determine the transfer timing and transfer path of the computing tasks between different cooling air ducts.

[0090] The system uses the temperature decay characteristics of the cooling duct to develop a strategy for migrating computing tasks between cooling ducts. For the first type of "high-heat" tasks, the system tends to schedule them to cooling ducts with fast temperature decay to accelerate heat dissipation; while for the second type of "low-heat" tasks, the system tends to schedule them to cooling ducts with slow temperature decay to balance the heat dissipation burden of each duct. By establishing task migration rules, the system can continuously optimize task scheduling plans under dynamically changing heat dissipation conditions, avoiding local overheating while taking into account overall heat dissipation efficiency.

[0091] In order to generate task migration rules, the system needs to evaluate the temperature decay capacity of each heat dissipation duct, that is, the speed at which the duct temperature drops after the heat generation task in the duct stops. The system can regularly perform standardized diagnostic procedures in each duct, such as high-load stress test, low-load sleep test, etc., to collect the decay curve of the duct temperature. By performing parameter identification on the decay curve, the system can estimate the temperature decay rate of each duct and form the heat dissipation capacity spectrum of the duct. On this basis, the system designs two sets of migration rules for the first and second types of tasks respectively. For the first type of tasks, the system monitors the temperature level of the duct in real time, and dynamically schedules the task to the duct with the fastest temperature decay and the lowest current temperature to achieve rapid heat dissipation. For the second type of tasks, the system adopts a round-robin scheduling strategy so that each duct undertakes a similar number of low-heat tasks to maintain overall heat dissipation balance.

[0092] The actual temperature decay rate of the heat dissipation duct may be affected by factors such as the airflow organization and component layout in the chassis, and these factors may change during the use of the host, causing the solidified task migration rules to lose adaptability. To address this problem, on the one hand, the system needs to adopt a reasonable internal layout during the host design phase, and use airflow simulation and other means to optimize the duct to improve the stability of the heat dissipation performance; on the other hand, the system can introduce an adaptive task migration mechanism to dynamically adjust the threshold and strategy parameters of task migration based on the real-time feedback of the duct temperature and task execution. For example, if the system finds that the actual heat dissipation capacity of a certain duct continues to be lower than expected, it can automatically increase the temperature threshold of the duct to delay the migration of high-heat tasks to it.

[0093] S105. When it is detected that the temperature in any heat dissipation duct is greater than the upper temperature threshold, the first type of computing tasks being executed in the heat dissipation duct are migrated to the target heat dissipation duct whose temperature is lower than the preset temperature threshold; the second type of computing tasks being executed in the target heat dissipation duct are migrated to the heat dissipation duct.

[0094] In this step, the system dynamically adjusts the execution position of the computing task based on the real-time temperature feedback of the cooling duct to achieve overheating protection for the host. When the temperature of a cooling duct exceeds the safety upper limit, the system immediately transfers the first-class high-heat tasks in the duct to the target duct with a lower temperature to prevent local temperature from getting out of control. At the same time, in order to balance the heat dissipation pressure of each duct, the system migrates some of the second-class low-heat tasks originally executed by the target duct to the overheating duct, taking over the execution of the original high-heat tasks and maintaining the basic working state of the duct. Through this dynamic task migration scheduling, the system can quickly respond to heat dissipation crises, suppress overheating risks, and ensure the stable operation of the host.

[0095] In order to achieve temperature-triggered task migration, the system needs to deploy a real-time temperature monitoring and early warning mechanism. The system uses a temperature sensor network distributed at key locations of the heat dissipation duct to collect duct temperature data at a high sampling frequency and transmit the data back to the central control unit in real time. The control unit continuously evaluates the duct temperature based on the set upper temperature threshold. Once it is found that the temperature of a duct exceeds the threshold, the control unit immediately initiates the predetermined task migration emergency plan: suspend the current first-class task of the duct, package its status data, and transmit it to the target duct with the lowest temperature through a high-speed data channel to resume the execution of the task; at the same time, the control unit selects several low-heat tasks from the target duct and transfers them to the overheated duct to maintain the power consumption balance of the duct. The entire task migration process can be completed in milliseconds, minimizing the impact on business continuity.

[0096] In the above embodiment, by constructing multiple heat dissipation ducts inside the host and obtaining the temperature response rate and temperature attenuation rate, the heat dissipation characteristics of different areas can be accurately grasped. By classifying computing tasks based on the temperature response rate, it is possible to identify the types of tasks that have a greater and lesser impact on temperature. Task migration rules are formulated in combination with the temperature attenuation rate. When a temperature anomaly is detected, the high-temperature generation task is migrated to the low-temperature area, and the low-temperature generation task in the low-temperature area is replaced to the original area, forming a virtuous cycle of temperature balance, improving the accuracy of controlling the transfer and accumulation of heat inside the device, and thereby improving the accuracy of temperature control management.

[0097] The above temperature balancing control method based on heat dissipation duct construction and task classification migration can achieve preliminary balance of the host's internal temperature. However, in the actual task migration process, the migration of different tasks will cause different degrees of fluctuations in system performance. If these performance fluctuations are not carefully managed, it may affect the stable operation of the system. Therefore, it is necessary to further consider the impact of task migration on system performance on the basis of temperature balancing control and formulate corresponding optimization strategies. Figure 2 , a task migration optimization method considering performance fluctuations in an embodiment of the present application is described:

[0098] See also Figure 2 , which is a flow chart of a task migration optimization method considering performance fluctuations in an embodiment of the present application.

[0099] S201, obtaining performance parameters of each functional component during the migration process of the computing task, and determining the fluctuation time interval and fluctuation amplitude of the performance parameters;

[0100] The system combines the first type of computing tasks and the second type of computing tasks into a number of computing task units according to performance parameters. Each computing task unit contains multiple computing tasks that occupy different types of computing resources.

[0101] The purpose of this step is to obtain the performance of each functional component of the system during the computing task migration process, so as to optimize the task migration plan based on the fluctuation of performance parameters. Among them, performance parameters can include indicators of multiple dimensions such as CPU utilization, memory utilization, I / O throughput, etc. The system can fully understand the impact of computing task migration on system performance by monitoring these performance parameters in real time.

[0102] In specific implementations, the system can use a variety of methods to obtain performance parameters, such as real-time collection of performance data of each functional component through hardware monitoring tools, performance counters provided by the operating system, etc. At the same time, in order to accurately evaluate the fluctuation of performance parameters, the system also needs to perform statistical analysis on the collected performance data and calculate the fluctuation amplitude of the performance parameter within a time interval, that is, the difference between the maximum and minimum values ​​of the performance parameter within the interval.

[0103] S202, combining the first type of computing tasks and the second type of computing tasks into a plurality of computing task units according to the performance parameters;

[0104] The system sets the migration execution timing of each computing task unit based on the fluctuation time interval, and the migration execution timing makes the migration time interval of adjacent computing task units greater than a preset time threshold.

[0105] The purpose of this step is to combine different types of computing tasks into several computing task units according to the similarity of performance parameters, so that in the subsequent task migration process, a more refined migration strategy can be formulated according to the performance characteristics of the computing task units. Among them, the first type of computing tasks and the second type of computing tasks occupy different types of computing resources, such as CPU-intensive tasks and I / O-intensive tasks, and their impact on system performance is quite different, so they need to be handled separately.

[0106] In the specific implementation, the system can use clustering algorithms to aggregate computing tasks with similar performance parameters into the same computing task unit. Commonly used clustering algorithms include K-means, hierarchical clustering, etc. At the same time, in order to ensure the rationality of the clustering results, the system also needs to tune the parameters in the clustering process, such as the distance measurement method of clustering, the number of clusters of clustering, etc., to obtain the best clustering effect.

[0107] S203, setting the migration execution sequence of each computing task unit based on the fluctuation time interval;

[0108] The system sets the migration execution sequence of each computing task unit based on the fluctuation time interval. The migration execution sequence makes the migration time interval of adjacent computing task units greater than the preset time threshold. Specifically: the performance recovery time of each computing task unit in the historical migration process is obtained. The performance recovery time is the time required for the performance parameters to recover to the preset benchmark level after the computing task migration is completed;

[0109] The computing task units are divided into several migration batches according to the performance recovery time, and the sum of the performance recovery time of the computing task units in each migration batch is less than the fluctuation time interval;

[0110] Determine the execution order of each migration batch so that the migration time interval between adjacent migration batches is not less than the maximum performance recovery time in the migration batch;

[0111] In each migration batch, migration time points are assigned to computing task units in the order of performance fluctuation from small to large;

[0112] When the actual performance recovery time of any computing task unit is greater than the historical performance recovery time, all migration time points after the migration batch will be postponed;

[0113] Generate a migration execution sequence including the migration time points of each computing task unit.

[0114] The purpose of this step is to reasonably arrange the migration execution order and time points of each computing task unit according to the fluctuation time interval of the performance parameters, and try to avoid task migration during the period of drastic performance fluctuations to reduce the impact of migration on system performance. Among them, the fluctuation time interval of the performance parameters reflects the drastic changes in system performance over a period of time, which is an important basis for formulating the migration execution sequence.

[0115] In the specific implementation, the system first needs to obtain the performance recovery time of each computing task unit during the historical migration process, that is, the time required for the performance parameters to return to the pre-migration level after the computing task migration is completed. Then, the system divides the computing task units into several migration batches according to the performance recovery time. The sum of the performance recovery time of the computing task units in each migration batch needs to be less than the fluctuation time interval to ensure that all computing task units can complete the migration and restore performance within a fluctuation time interval. Next, the system determines the execution order of each migration batch, and ensures that the migration time interval between adjacent migration batches is not less than the maximum performance recovery time in the migration batch, to avoid continuous task migration causing continuous impact on system performance. Finally, the system assigns specific migration time points to the computing task units in each migration batch in the order of performance fluctuation amplitude from small to large, and dynamically adjusts the migration time points to adapt to the actual performance recovery situation, and finally generates a complete migration execution sequence.

[0116] In the process of formulating the migration execution sequence, the performance recovery time of some computing task units may be much longer than that of other units, or the performance recovery time estimation may be inaccurate, resulting in the inability to execute task migration according to the scheduled sequence. In response to this situation, the system can dynamically adjust the division and execution order of migration batches, arrange computing task units with particularly long performance recovery times separately in the last migration batch, or increase the time interval between adjacent migration batches to ensure the smoothness of the migration process. In addition, the system can also monitor the performance recovery of computing task units in real time during the migration execution process, and dynamically adjust the migration time points of subsequent computing task units according to the actual performance recovery time to address the problem of inaccurate performance recovery time estimation.

[0117] S204, in the process of executing the migration execution sequence, alternately migrate the computing task units whose performance fluctuation amplitude is greater than the average fluctuation amplitude and the computing task units whose performance fluctuation amplitude is less than the average fluctuation amplitude;

[0118] The purpose of this step is to further reduce the impact of task migration on system performance by alternately migrating computing task units with different performance fluctuation characteristics during the execution of the migration sequence. The performance fluctuation amplitude reflects the degree of impact of the computing task unit on system performance during the migration process. The larger the fluctuation amplitude, the more severe the system performance fluctuation during the migration of the computing task unit.

[0119] In the specific implementation, in each migration batch, the system first migrates the computing task units whose performance fluctuation amplitude is less than or equal to the average fluctuation amplitude, and then migrates the computing task units whose performance fluctuation amplitude is greater than the average fluctuation amplitude. This alternating migration method can avoid the continuous migration of multiple computing task units with drastic performance fluctuations, which will cause continuous and drastic fluctuations in system performance. At the same time, each time the migration performance fluctuation amplitude size type is switched, the system also needs to reserve sufficient time interval to ensure that the system performance can be restored to a stable state after the migration of the previous type of computing task units is completed.

[0120] S205: When the fluctuation amplitude of the performance parameter exceeds a preset time threshold, the migration time of subsequent computing task units is postponed.

[0121] The purpose of this step is to postpone the migration time of subsequent calculation task units when the system performance is monitored to have drastic fluctuations during the task migration process, so as to avoid continuous task migration when the system performance is unstable, which may lead to further deterioration of system performance. The preset time threshold is an upper limit of performance fluctuation set according to factors such as the system's performance tolerance and the urgency of task migration. When the actual performance fluctuation amplitude exceeds the threshold, it means that the system performance has been significantly affected and it is necessary to suspend task migration and wait for the system performance to return to normal levels before continuing.

[0122] In the specific implementation, the system monitors the fluctuation amplitude of the performance parameters in the current time period in real time during the execution of the migration sequence, and compares it with the preset time threshold. Once it is found that the fluctuation amplitude of the performance parameter exceeds the preset time threshold, the system immediately suspends the migration of the current computing task unit, and postpones the migration time of all subsequent computing task units in the migration sequence until the system performance returns to a stable state. When postponing the migration time, the system needs to re-evaluate the performance recovery time of each computing task unit, and adjust the division and execution order of the migration batches accordingly to ensure that after the system performance is restored, the remaining computing task units can complete the migration in the shortest time.

[0123] In the above embodiment, by obtaining the performance parameter fluctuation characteristics during the task migration process and combining different types of computing tasks into computing task units, the system achieves balanced utilization of computing resources. The migration execution sequence set based on the performance fluctuation characteristics ensures the smooth progress of the task migration process. The strategy of alternately migrating computing task units with different performance fluctuation amplitudes smoothes the overall performance fluctuations. When performance fluctuations beyond expectations occur, by postponing the migration time of subsequent tasks, a buffer time for performance recovery is reserved for the system, while achieving temperature balance, maintaining the stability of system performance, and improving the stability of the system during temperature control.

[0124] Furthermore, in another embodiment, after generating the migration execution sequence including the migration time points of each computing task unit, the system further includes:

[0125] Calculate the migration duration of each computing task unit under the migration execution sequence, where the migration duration is the time required for the computing task to start migration and complete migration;

[0126] Determine the migration priority of each computing task unit according to the migration duration and performance recovery time;

[0127] When it is detected that the temperature change rate of the heat dissipation air duct exceeds a preset temperature threshold, the migration order of the computing task units with the same migration time point is adjusted, specifically including:

[0128] Obtain the resource occupancy type and occupancy ratio of each computing task unit;

[0129] Calculate the influence coefficient of each computing task unit on the temperature of the heat dissipation duct;

[0130] Sort the computing task units with the same migration time point according to the influence coefficient;

[0131] Arrange the computing task units whose influence coefficient is greater than a preset influence coefficient threshold to be executed in a time period when the temperature is lower than the target temperature;

[0132] When the temperature change rate drops below a preset temperature threshold, the migration sequence is restored.

[0133] First, the system calculates the migration duration required for each computing task unit from the start of migration to the completion of migration, and the performance recovery time required for the performance parameters to return to normal levels after the migration is completed. Based on these two time parameters, the system can evaluate the migration urgency of each computing task unit and the degree of impact on system performance, and determine their migration priority. For example, for computing task units with short migration duration and long performance recovery time, the system can give them a higher migration priority to complete the migration of these task units as soon as possible and reduce the impact time on system performance.

[0134] Secondly, the system monitors the temperature changes of the heat dissipation duct in real time during the task migration process. When the temperature change rate exceeds the preset temperature threshold, it means that the heat dissipation efficiency of the current heat dissipation duct is low, and the migration order of the computing task units needs to be optimized and adjusted to reduce the heat dissipation pressure. Specifically, the system first obtains the resource occupancy type (such as CPU intensive, I / O intensive, etc.) and occupancy ratio of each computing task unit, and then calculates the temperature influence coefficient of each computing task unit according to the degree of influence of different resource types on the temperature of the heat dissipation duct. Next, the system sorts the computing task units with the same migration time point according to the temperature influence coefficient. The larger the influence coefficient, the more heat the task unit generates during operation, and the greater the impact on the temperature of the heat dissipation duct.

[0135] Then, according to the real-time temperature of the cooling duct, the system will try to arrange the computing task units with a temperature impact coefficient greater than the preset threshold to be executed in a time period with lower temperature, and arrange the task units with a smaller temperature impact coefficient to be executed in a time period with higher temperature, thereby ensuring the progress of task migration while reducing the impact of high temperature impact coefficient task units on cooling efficiency. When the temperature of the cooling duct drops and the temperature change rate drops below the preset temperature threshold again, the system restores the original migration order.

[0136] In the above embodiment, the migration priority of the task unit is determined by calculating the migration duration and performance recovery time, and the system can quantitatively evaluate the impact of each task migration on the overall performance. When it is detected that the temperature change rate of the heat dissipation duct exceeds the threshold, the migration order of the tasks at the same time point is adjusted in a targeted manner, so that the system can perform local optimization on the basis of maintaining the original migration plan framework, and can flexibly adjust the task execution order when facing temperature anomalies, and optimize the temperature control effect without affecting the overall migration plan. Through the comprehensive consideration of the two key indicators of migration duration and performance recovery time, the system can achieve a dynamic balance between performance and temperature during the task migration process, improve the accuracy and timeliness of temperature control, and ensure a smooth transition of system performance.

[0137] The following describes the system in the embodiment of the present invention from the perspective of hardware processing. Figure 3 , which is a schematic diagram of the physical device structure of a host power consumption and temperature balance control system provided in an embodiment of the present application.

[0138] It should be noted that Figure 3 The structure of the system shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0139] like Figure 3As shown, the system includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 302 or the program loaded from the storage part 308 to the random access memory (RAM) 303, such as executing the method in the above embodiment. In RAM 303, various programs and data required for system operation are also stored. CPU 301, ROM 302 and RAM 303 are connected to each other through a bus 304. Input / output (I / O) interface 305 is also connected to bus 304.

[0140] The following components are connected to the I / O interface 305: an input section 306 including a camera, an infrared sensor, etc.; an output section 307 including a liquid crystal display (LCD) and a speaker, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. A removable medium 311, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 310 as needed so that a computer program read therefrom is installed into the storage section 308 as needed.

[0141] In particular, according to an embodiment of the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through the communication part 309, and / or installed from a removable medium 311. When the computer program is executed by the central processing unit (CPU) 301, various functions defined in the present invention are performed.

[0142] It should be noted that the computer-readable medium shown in the embodiment of the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, device or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable computer program. Such a propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing.

[0143] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. Among them, each box in the flowchart or block diagram can represent a module, a program segment, or a part of the code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0144] As another aspect, the present invention further provides a computer-readable storage medium, which may be included in the system described in the above embodiment; or may exist independently without being assembled into the system. The above storage medium carries one or more computer programs, and when the above one or more computer programs are executed by a processor of a system, the system implements the method provided in the above embodiment.

[0145] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0146] As used in the above embodiments, the term "when..." may be interpreted as "if..." or "after..." or "in response to determining..." or "in response to detecting...", depending on the context. Similarly, the phrases "upon determining..." or "if (the stated condition or event) is detected" may be interpreted as "if determining..." or "in response to determining..." or "upon detecting (the stated condition or event)" or "in response to detecting (the stated condition or event)", depending on the context.

[0147] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk), etc.

[0148] Those skilled in the art can understand that to implement all or part of the processes in the above-mentioned embodiments, the processes can be completed by computer programs to instruct related hardware, and the programs can be stored in computer-readable storage media. When the programs are executed, they can include the processes of the above-mentioned method embodiments. The aforementioned storage media include: ROM or random access memory RAM, magnetic disk or optical disk and other media that can store program codes.

Claims

1. A method for balancing host power consumption and temperature, characterized in that: include: Determine a temperature data sequence based on the air flow data inside the host collected in real time by the temperature sensor; Constructing a temperature field distribution model inside the host according to the temperature data sequence; determining an optimal connection path between functional components using a minimum spanning tree algorithm based on the temperature field distribution model; combining the functional components having the same optimal connection path into a heat dissipation duct to obtain a plurality of heat dissipation ducts; Collecting temperature data in each heat dissipation duct, and calculating the temperature response rate and temperature decay rate of each heat dissipation duct respectively, wherein the temperature response rate represents the speed of temperature rise in the heat dissipation duct, and the temperature decay rate represents the speed of temperature drop in the heat dissipation duct; Constructing a resource consumption feature vector of a computing task, wherein the feature vector includes processor occupancy, memory occupancy, and input / output frequency; Performing cluster analysis on the characteristic vectors by using an adaptive clustering algorithm to obtain several cluster centers; calculating the temperature response characteristic curve corresponding to each cluster center; Extracting key inflection points of the temperature response characteristic curve, and determining a critical interval of temperature rise based on the key inflection points; Dividing the computing tasks into first-category computing tasks and second-category computing tasks according to the critical interval; Establishing migration rules for the first type of computing tasks and the second type of computing tasks according to the temperature decay rate of each heat dissipation air duct, wherein the migration rules are used to determine the transfer timing and transfer path of the computing tasks between different heat dissipation air ducts; When it is detected that the temperature corresponding to the temperature data of any heat dissipation air duct is greater than the upper temperature threshold, the first type of computing task of the heat dissipation air duct whose temperature is greater than the upper temperature threshold is suspended, and the first type of computing task of the heat dissipation air duct whose temperature is greater than the upper temperature threshold is transmitted to the target heat dissipation air duct with the lowest temperature through a high-speed data channel, so that the target heat dissipation air duct executes the first type of computing task; Transferring the second type of computing tasks in the target heat dissipation duct to the heat dissipation duct whose temperature is greater than the upper temperature threshold and executing the second type of computing tasks; Acquire the performance parameters of each of the functional components during the migration of the computing task, and determine the fluctuation time interval and fluctuation amplitude of the performance parameters; Combining the first type of computing tasks and the second type of computing tasks into a plurality of computing task units according to the performance parameters, wherein the plurality of computing tasks included in each of the computing task units occupy different types of computing resources respectively; Setting a migration execution sequence of each of the computing task units based on the fluctuation time interval, wherein the migration execution sequence makes the migration time interval between adjacent computing task units greater than a preset time threshold; In the process of executing the migration execution sequence, the computing task units whose performance fluctuation amplitude is greater than the average value of the fluctuation amplitude and the computing task units whose performance fluctuation amplitude is less than the average value of the fluctuation amplitude are alternately migrated; When the fluctuation amplitude of the performance parameter exceeds the preset time threshold, the migration time of the subsequent computing task unit is postponed.

2. The method according to claim 1, characterized in that The step of setting the migration execution sequence of each computing task unit based on the fluctuation time interval specifically includes: Obtaining the performance recovery time of each computing task unit during the historical migration process, wherein the performance recovery time is the time required for the performance parameter to recover to a preset benchmark level after the computing task migration is completed; Dividing the computing task units into a plurality of migration batches according to the performance recovery time, wherein the sum of the performance recovery time of the computing task units in each migration batch is less than the fluctuation time interval; Determining the execution order of each migration batch so that the migration time interval between adjacent migration batches is not less than the maximum performance recovery time in the migration batch; Allocating migration time points to the computing task units in each migration batch according to the performance fluctuation amplitude from small to large order; When the actual performance recovery time of any of the computing task units is greater than the historical performance recovery time, all migration time points after the migration batch are postponed; A migration execution sequence including migration time points of each computing task unit is generated.

3. The method according to claim 2, characterized in that After generating the migration execution sequence including the migration time points of each computing task unit, the method further includes: Calculate the migration duration of each computing task unit under the migration execution sequence, where the migration duration is the time required for the computing task to start migration and complete migration; Determining the migration priority of each of the computing task units according to the migration duration and the performance recovery time; When it is detected that the temperature change rate of the heat dissipation air duct exceeds a preset temperature threshold, the migration order of the computing task units having the same migration time point is adjusted.

4. The method according to claim 3, characterized in that The step of adjusting the migration order of computing task units having the same migration time point specifically includes: Obtaining the resource occupancy type and occupancy ratio of each of the computing task units; Calculating the influence coefficient of each computing task unit on the temperature of the heat dissipation duct; Sorting the computing task units having the same migration time point according to the influence coefficient; Arrange the computing task unit whose influence coefficient is greater than a preset influence coefficient threshold to be executed in a time period when the temperature is lower than the target temperature; When the temperature change rate decreases below the preset temperature threshold, the migration sequence is restored.

5. A host power consumption and temperature balance control system, characterized in that: The system comprises: One or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to cause the system to execute the method as described in any one of claims 1-4.

6. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on a system, the system is caused to execute the method according to any one of claims 1 to 4.

7. A computer program product, characterized in that When the computer program product is run on a system, the system is caused to execute the method according to any one of claims 1 to 4.

Images