A fan speed control method and device based on hard drive read / write performance monitoring

Abstract

This application provides a fan speed control method and apparatus based on hard disk read / write performance monitoring, belonging to the field of server hard disk heat dissipation. The method includes: acquiring multiple real-time read / write performance parameters of the hard disk; calculating the ratio of each real-time read / write performance parameter to the current maximum allowable performance parameter of the hard disk; calculating a weighted sum of the ratios to obtain the dynamic load score of the hard disk; calculating a speed control value based on the nonlinear relationship between the dynamic load score and the fan speed, wherein the nonlinear relationship is a piecewise linear function, and the dynamic load score is a target piecewise linear function determined from multiple piecewise linear functions; and controlling the fan rotation according to the speed control value. The fan speed control method and apparatus based on hard disk read / write performance monitoring provided in this application can solve the problems of hard disk performance degradation and ineffective fan energy consumption caused by traditional temperature-controlled speed control strategies.

Description

Technical Field

[0001] This application relates to the field of server hard drive heat dissipation technology, and in particular to a fan speed control method and device based on hard drive read / write performance monitoring. Background Technology

[0002] In servers, storage arrays, and other devices that use hard drives as their core storage component, the hard drive's operating temperature is directly related to its operational stability, lifespan, and performance. When performing data read / write operations, the computational load of the internal chips and the frequency of the read / write head movement change synchronously with the load intensity, leading to fluctuations in heat generation. Under high-load read / write scenarios (such as large-scale database queries and batch file transfers), the hard drive's heat generation increases significantly. If heat dissipation is insufficient and the temperature exceeds the safe operating limit (typically 55-60℃), it will trigger the hardware frequency reduction protection mechanism, causing a decrease in data read / write speed. Prolonged exposure to high temperatures also accelerates the aging of critical components such as the read / write heads and platters, increasing the risk of hard drive failure. Conversely, if the fan always runs at high speed to ensure heat dissipation, it generates additional energy consumption and noise, especially under low-load scenarios, where this excessive heat dissipation leads to resource waste. Therefore, how to dynamically adjust the fan speed according to the actual operating status of the hard drive, achieving a balance between effective heat dissipation and energy saving and noise reduction, has become a pressing technical problem to be solved in the server hard drive field.

[0003] Currently, the mainstream fan speed control solutions in the industry mainly revolve around two logic designs: fixed speed or single temperature trigger. Among them, the fixed speed solution uses preset fan operating levels (such as low load level, high load level) or directly sets a constant speed. The control logic is simple and easy to implement, but it cannot adapt to the dynamically changing load requirements of hard drives. At low loads, the fan runs at high speed, resulting in wasted energy, while at high loads, it may not be able to dissipate heat in time due to insufficient speed. The single temperature trigger solution uses the hard drive's built-in sensor to collect real-time temperature. When the temperature exceeds a preset threshold, the speed is increased, and when it is below the threshold, the speed is decreased. Although it achieves a certain degree of dynamic adjustment, temperature changes often lag behind load changes (for example, when the load suddenly increases, the heat generation increases first, and the temperature takes 10-30 seconds to rise significantly). It is prone to problems with untimely heat dissipation response and cannot distinguish between different scenarios such as low load with high ambient temperature and high load with low ambient temperature, which may lead to a disconnect between speed control decisions and actual heat dissipation requirements.

[0004] Furthermore, existing solutions lack adaptability in scenarios involving collaborative cooling of multiple hard drives. Server chassis typically house multiple hard drives, and the cooling space of each drive is interconnected. High load on one hard drive can passively raise the temperature of surrounding hard drives. Existing solutions do not consider the mutual influence of heat dissipation between hard drives, easily leading to problems such as excessively high temperatures in some hard drives and low overall cooling efficiency. At the same time, neither fixed speed nor single temperature-triggered solutions can completely solve the problem of performance degradation and inefficient energy consumption coexisting. Summary of the Invention

[0005] In view of this, this application provides a fan speed control method and device based on hard disk read / write performance monitoring to solve the problems of hard disk performance degradation and ineffective fan power consumption caused by traditional temperature control speed control strategies.

[0006] Specifically, this application is implemented through the following technical solution:

[0007] The first aspect of this application provides a fan speed control method based on hard disk read / write performance monitoring, the method comprising:

[0008] Obtain multiple real-time read and write performance parameters of the hard drive;

[0009] Calculate the ratio of the real-time read / write performance parameters of each hard drive to the current maximum allowable performance parameters of the hard drive, wherein the current maximum allowable performance parameters of the hard drive are calculated in real time based on the space layout of the hard drive and the total computational load of data read / write requests;

[0010] Calculate the weighted sum of each ratio to obtain the dynamic load score of the hard disk;

[0011] The speed control value is calculated based on the nonlinear relationship between the dynamic load score and the fan speed, wherein the nonlinear relationship is a piecewise linear function, and the dynamic load score is determined from multiple piecewise linear functions to identify the target piecewise linear function;

[0012] The fan is controlled to rotate according to the speed control value.

[0013] The second aspect of this application provides a fan speed control device based on hard disk read / write performance monitoring, the device comprising an acquisition module, a calculation module, and a processing module;

[0014] The acquisition module is used to acquire multiple real-time read and write performance parameters of the hard drive;

[0015] The calculation module is used to calculate the ratio of the real-time read / write performance parameters of each hard drive to the current maximum allowable performance parameters of the hard drive, wherein the current maximum allowable performance parameters of the hard drive are calculated in real time by the space layout of the hard drive and the total amount of data read / write requests.

[0016] The calculation module is also used to calculate the weighted sum of the ratios to obtain the dynamic load score of the hard disk;

[0017] The calculation module is also used to calculate the speed control value based on the nonlinear relationship between the dynamic load score and the fan speed, wherein the nonlinear relationship is a piecewise linear function, and the dynamic load score is determined from multiple piecewise linear functions to identify the target piecewise linear function;

[0018] The processing module is used to control the fan rotation according to the speed control value.

[0019] This application provides a fan speed control method and device based on hard drive read / write performance monitoring. It offers a fan speed control scheme with hard drive dynamic load as the core and piecewise linear functions as the control logic. This scheme can accurately correlate complex hard drive loads with fan speeds by real-time acquisition of hard drive read / write performance parameters and dynamic calculation of load scores. It provides a precise and dynamic fan speed control process based on parameter acquisition, ratio calculation, load quantification, and speed adjustment, supported by the hard drive's current maximum allowable performance parameters and piecewise linear functions, and grounded in the hard drive's real-time load adaptability. Specifically, by acquiring multiple real-time read / write performance parameters and calculating the ratios of these parameters to the current maximum allowable performance parameters, and combining this with the hard drive's space layout and actual read / write request volume to determine a dynamic benchmark, it achieves accurate acquisition and standardized processing of hard drive load data. Real-time read / write performance parameters provide multi-dimensional raw data for load assessment, while parameter ratio calculation normalizes data from different dimensions, avoiding load misjudgments caused by differences in parameter units. By calculating the weighted sum of each ratio to obtain a dynamic load score, and then calculating the speed control value based on a piecewise linear function, a direct correlation is established between the overall hard drive load and the fan speed. The dynamic load score quantifies the actual working intensity of the hard drive, while the piecewise linear function avoids the blindness of traditional fixed speed or single temperature trigger strategies through differentiated speed adjustment rules (low speed under low load, high speed under high load), improving the targeting and efficiency of fan speed adjustment, and allowing users to intuitively control the heat dissipation rhythm based on load changes.

[0020] By using a design where the maximum allowable performance parameters of the hard drive are calculated in real time based on the hard drive space layout and the total computational load of data read and write requests, users do not need to rely on fixed theoretical hardware values ​​to set evaluation standards. The system can automatically determine the benchmark by combining the physical capabilities of the hard drive with actual needs, avoiding the problems of underestimating the load when the hardware capability is excessive but the demand is low, and overestimating the load when the demand exceeds the hardware capability. This allows the dynamic load score to truly reflect the current operating pressure of the hard drive. The design of determining the target piecewise linear function from multiple piecewise linear functions through the dynamic load score achieves flexible adaptation of speed adjustment rules and compatibility coverage of different scenarios. The system can preset multiple sets of piecewise linear functions for different hard drive models and different usage scenarios. The dynamic load score can automatically match the corresponding function to calculate the speed, avoiding the limitations of a single speed adjustment rule that cannot adapt to diverse hardware and scenarios. This allows the fan speed adjustment to meet the powerful heat dissipation requirements of high-load scenarios while also taking into account the energy saving and noise reduction goals of low-load scenarios, further improving the practicality and adaptability of the solution. Attached Figure Description

[0021] Figure 1 A flowchart of an embodiment of the fan speed control method based on hard disk read / write performance monitoring provided in this application;

[0022] Figure 2 This is a schematic diagram of the second embodiment of the fan speed control device based on hard disk read / write performance monitoring provided in this application. Detailed Implementation

[0023] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application.

[0024] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used herein are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

[0025] It should be understood that although the terms first, second, third, etc., may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0026] The following specific embodiments are given to illustrate the technical solution of this application in detail.

[0027] Example 1

[0028] Figure 1 This is a flowchart of an embodiment of the fan speed control method based on hard disk read / write performance monitoring provided in this application. Please refer to... Figure 1 The method provided in this embodiment may include:

[0029] S101: Obtain multiple real-time read / write performance parameters of the hard drive.

[0030] It's important to note that when executing fan speed control based on hard drive read / write performance monitoring, the first step is to acquire real-time hard drive read / write performance parameters. This is the foundation for subsequent precise speed adjustment. Real-time read / write performance parameters refer to key indicators that dynamically reflect the current data read / write operation status and efficiency of the hard drive. This data needs to be continuously acquired at a preset collection cycle (e.g., 1 second / time or 5 seconds / time, which can be flexibly set according to the hard drive's usage scenario) to ensure that the data matches the actual working status of the hard drive in real time.

[0031] In terms of parameter types, the real-time read / write performance parameters in this embodiment should at least include the hard disk data transfer rate. This parameter is an indispensable core indicator, which directly reflects the amount of data read / write that the hard disk can complete per unit time (usually in MB / s or GB / s), and is a key basis for measuring the hard disk load intensity. In addition to the core hard disk data transfer rate, it can also optionally include the number of input / output operations per second (IOPS) and seek latency. The number of input / output operations per second is used to characterize the frequency of read / write operations that the hard disk can complete per second (in times / second), while the seek latency is used to reflect the average time required for the hard disk head to move to a specified data sector (usually in milliseconds). Both can be used as auxiliary indicators to further improve the accuracy of judging the hard disk load status.

[0032] In practice, the system can interact with the underlying driver of the operating system through the hard drive controller interface (such as SATA, SAS or NVMe interface) to directly read the performance data collected in real time by the built-in sensors or controller of the hard drive. Alternatively, a third-party hardware monitoring tool (which needs to be compatible with the corresponding hard drive model and operating system version) can be used to collect and store the above real-time read and write performance parameters at a preset period to ensure that the data collection process does not affect the normal read and write operation of the hard drive and that the data transmission delay is controlled within a preset range (such as not exceeding 100 milliseconds) to avoid deviations in subsequent speed adjustment decisions due to data lag.

[0033] It's important to note that prioritizing hard drive data transfer rate as the core parameter, and selectively supplementing other performance indicators, is because in most server and storage array applications, hard drive data transfer rate has the strongest correlation with actual power consumption and heat generation. As the transfer rate increases, the computational load on the hard drive's internal chips and the frequency of read / write head activity increase simultaneously, leading to a rise in heat generation. The number of input / output operations per second and seek latency only significantly affect heat generation in specific scenarios (such as random read / write intensive tasks). Therefore, it's unnecessary to forcibly include all parameters. Focusing solely on the hard drive data transfer rate and flexibly selecting whether to supplement other indicators based on the actual application scenario ensures both efficient data collection and accurate data support for subsequent load analysis.

[0034] It should be noted that, in this embodiment, real-time read / write performance parameters include at least the number of input / output operations per second, hard disk data transfer rate, and seek latency.

[0035] S102. Calculate the ratio of the real-time read / write performance parameters of each hard disk to the current maximum allowable performance parameters of the hard disk, wherein the current maximum allowable performance parameters of the hard disk are calculated in real time by the space layout of the hard disk and the total amount of data read / write requests.

[0036] It's important to note that the maximum permissible performance parameter of a hard drive is not a fixed value, unlike the static theoretical maximum value stated in the hard drive's specifications. From an actual operational perspective, the performance ceiling a hard drive can achieve at any given moment is directly influenced by two key factors. First, the hard drive's physical structure, specifically its spatial layout, such as the physical division of different storage areas and whether it supports parallel read / write operations across multiple areas. These hardware-level designs directly determine the hardware capabilities of the hard drive during data transfer. Second, the system's immediate task demands on the hard drive—that is, the total computational load of data read / write requests from the operating system and applications at any given moment. This demand-side factor determines the actual workload the hard drive needs to handle. Due to the combined effect of these two factors, the maximum permissible performance parameter of the hard drive needs to be calculated based on real-time conditions. It is a dynamically changing baseline value, serving as a reference for subsequent calculations and more closely reflecting the hard drive's current actual operating state.

[0037] In other words, the hard drive's spatial layout determines its maximum parallel read / write capability (i.e., parallel read / write bandwidth) at the hardware level, which is the physical upper limit of hard drive performance. The total computational load of data read / write requests reflects the actual performance demands of the current system on the hard drive, representing the workload the hard drive needs to handle. Only by combining these two aspects to determine the maximum allowable performance parameters can we avoid load misjudgment caused by a single reference. If only the hardware upper limit is used as a reference, when the actual request volume is low, the load percentage will be underestimated due to an overly high parameter benchmark; if only the actual request volume is used as a reference, when the request volume exceeds the hardware capacity, the load percentage will be overestimated due to an overly high parameter benchmark.

[0038] As a benchmark for comparing various real-time read / write performance parameters, the reasonableness of the current maximum allowable performance parameter of the hard drive directly affects the accuracy of subsequent ratio calculations. In actual calculations, a maximum allowable value consistent with its physical dimension must be matched for different types of real-time read / write performance parameters, such as the number of input / output operations per second, hard drive data transfer rate, and seek latency. For example, the hard drive data transfer rate must correspond to the maximum allowable value in the transfer rate dimension, and the number of input / output operations per second must correspond to the maximum allowable value in the operation frequency dimension. Then, by dividing the real-time parameter value by the corresponding maximum allowable value, parameters from different dimensions are transformed into a uniformly comparable ratio. This operation provides a standardized data input basis for the subsequent weighted summation and calculation of the dynamic load score. Furthermore, the specific calculation process for the current maximum allowable performance parameter of the hard drive will be explained in detail below and will not be elaborated here.

[0039] It should be noted that before calculating the ratio of each hard drive's real-time read / write performance parameters to the hard drive's current maximum allowable performance parameters, the following steps are taken:

[0040] (1) Within a preset time window, determine the parallel read / write bandwidth of each storage area based on the physical space layout information of the hard disk.

[0041] It should be noted that a preset time window needs to be set first. The window duration can be adjusted inversely according to the fluctuation frequency of hard disk data read and write operations. That is, the higher the fluctuation frequency of hard disk data read and write operations (such as a sudden increase or decrease in the number of read and write requests in a short period of time), the shorter the window duration should be (e.g., set to 10 seconds) to quickly capture load changes and ensure data timeliness. If the fluctuation frequency is lower (such as maintaining a stable number of read and write requests for a long time), the window duration can be appropriately extended (e.g., set to 30 seconds) to reduce the system resource consumption caused by frequent calculations.

[0042] Specifically, determining the parallel read / write bandwidth of each storage area based on the physical space layout information of the hard drive includes: identifying and dividing independent storage areas that can be operated in parallel according to the physical space layout information; determining the theoretical read / write bandwidth of each independent storage area; and summing the theoretical read / write bandwidths of all independent storage areas that can be operated in parallel to determine the parallel read / write bandwidth. In specific implementation, the physical space layout information of the hard drive is extracted, including the hard drive's partitioning (such as the number and capacity allocation of primary and extended partitions), the distribution pattern of tracks and sectors (such as the difference in read / write speeds between different tracks), whether a RAID array is used (if so, the RAID level needs to be specified, such as the parallel read / write rules corresponding to RAID0, RAID5, etc.), and the hardware interface speed differences between different storage areas (such as the speed limits corresponding to the SATA and NVMe interfaces connecting the hard drive to the motherboard). Then, storage areas are divided according to physical independence and read / write correlation. Hardware units with independent read / write channels and capable of simultaneous data operations are divided into independent storage areas, such as different partitions of a single hard drive (if parallel read / write is supported) and a single member hard drive in a RAID array, all of which are considered independent storage areas. The theoretical read / write bandwidth of each independent storage area is calculated and determined based on the hardware characteristics of each area. For ordinary partitions, the theoretical read / write bandwidth is the smaller of the average read / write rate of the track where the partition is located and the upper limit of the hardware interface rate. For member hard drives in a RAID array, the theoretical read / write bandwidth of a single hard drive is calculated according to the parallel rules corresponding to the RAID level (e.g., in a RAID0 array, the theoretical bandwidth of a single hard drive is consistent with the upper limit of the interface rate). Finally, the theoretical read / write bandwidths of each independent storage area are summed to obtain the overall parallel read / write bandwidth of the hard drive. If some storage areas do not support parallel read / write (e.g., a partition only supports serial operation), only the bandwidth of the areas that support parallel read / write is included in the summation range to ensure that the final result reflects the maximum data transfer capacity that the hard drive hardware can simultaneously handle.

[0043] Parallel read / write bandwidth refers to the maximum total data transfer rate (usually measured in MB / s or GB / s) supported at the hardware level when multiple storage areas are read and written simultaneously. It directly reflects the upper limit of performance that the physical structure of the hard drive can support.

[0044] (2) Calculate the total computational load of data read and write requests within the preset time window.

[0045] Within the same preset time window as in step (1), the total computational load of all data read and write requests from the operating system and applications is simultaneously counted. The duration of the preset time window must meet real-time constraints, typically ranging from 1 second to 60 seconds. If the window duration is too short (e.g., less than 1 second), the complete request queue may not be accurately captured due to system I / O scheduling delays. If the window duration is too long (e.g., more than 60 seconds), the load assessment will be severely delayed and will not reflect the true instantaneous pressure on the hard drive.

[0046] Data read / write requests encompass both background data interactions of the operating system itself (such as system file updates and cache data read / write) and business data requests from various applications (such as database queries, file downloads / uploads, and data calls during video rendering). Total computational load is measured in data volume (units consistent with the parallel read / write bandwidth, such as MB or GB). This volume needs to be captured and accumulated in real-time using the operating system's process management interface, application logs, or dedicated data monitoring tools to obtain the total data volume within the entire time window. For example, within a 10-second time window, if the operating system's background read / write requests accumulate to 50MB, a database application's query and write requests accumulate to 200MB, and another file transfer application's requests accumulate to 150MB, then the total computational load within that window is 400MB.

[0047] (3) Compare the parallel read / write bandwidth with the total computational load of the data read / write requests, and take the smaller value of the two as the current maximum allowable performance parameter of the hard disk at the current moment.

[0048] After completing the above two steps, the two values ​​need to be compared, and the smaller value should be taken as the current maximum allowable performance parameter. Specifically, the actual maximum performance output of the hard drive is limited by both the hardware's capacity and the actual demand. If the parallel read / write bandwidth (hardware limit) is greater than the total computational load of data read / write requests (actual demand), it means that the current hardware capability is sufficient to meet the demand, and the hard drive does not need to run at full capacity. In this case, the actual demand (total computational load) should be used as the maximum allowable performance parameter to avoid underestimating the actual load due to calculation based on the hardware limit (e.g., if the hardware limit is 500MB / s and the actual demand is 200MB / s, use 200MB / s as the baseline). If the parallel read / write bandwidth is less than the total computational load of data read / write requests, it means that the current demand exceeds the hardware capability, and the hard drive can only operate at the hardware limit at most. In this case, the parallel read / write bandwidth should be used as the maximum allowable performance parameter to avoid overestimating the load due to calculation based on the actual demand (e.g., if the hardware limit is 300MB / s and the actual demand is 400MB / s, use 300MB / s as the baseline). In this way, by taking the smaller value, the final maximum allowable performance parameter can match the actual working capacity of the hard drive and meet the actual needs, providing an accurate and reasonable benchmark for the subsequent calculation of the ratio of various real-time performance parameters.

[0049] Furthermore, it should be noted that calculating the ratio of each hard drive's real-time read / write performance parameters to the hard drive's current maximum allowable performance parameters includes:

[0050] (1) Based on the determined current maximum allowable performance parameters, determine the maximum allowable values ​​associated with the number of input / output operations per second, the hard disk data transfer rate and the seek latency, respectively.

[0051] It should be noted that the current maximum allowable performance parameter is essentially a bandwidth benchmark in terms of data transfer rate (e.g., MB / s). Since the real-time read / write performance parameters to be compared contain different physical dimensions, the maximum allowable value for each dimension needs to be determined based on this benchmark and using preset rules for input / output operations per second (IOPS) and seek latency.

[0052] Specifically, firstly, the maximum allowable value for hard drive data transfer rate can be directly adopted from the current maximum allowable performance parameter, as they share the same physical dimensions. Secondly, the maximum allowable value for the number of input / output operations per second needs to be calculated. This calculation involves dividing the current maximum allowable performance parameter by a preset typical data block size. This data block size can be the hard drive's default read / write block size or the average data volume per request obtained from historical I / O operation statistics. This conversion transforms the bandwidth benchmark into an operation frequency benchmark. Finally, for seek latency, a lower value indicates better performance. Its maximum allowable value is a latency threshold, representing the worst possible response time allowed to maintain stable hard drive performance under current load and cooling conditions. This latency threshold can be obtained by establishing a correlation model with the current maximum allowable performance parameter through offline experiments. For example, in high-bandwidth load scenarios, frequent head seek operations significantly increase latency, allowing the model to output a corresponding, more stringent latency upper limit.

[0053] To better understand, let's illustrate with an example. Assume the current maximum allowable performance parameter is 100MB / s, and the typical data block size set by the system is 1MB. Then, the comparable maximum allowable value for hard disk data transfer rate is 100MB / s; the comparable maximum allowable value for the number of input / output operations per second is 100MB / s ÷ 1MB / operation = 100 operations / second; simultaneously, according to the correlation model, under the current bandwidth pressure of 100MB / s, the comparable maximum allowable value for seek latency is set to 20 milliseconds. In this way, comparable maximum allowable values ​​are matched for real-time performance parameters of different dimensions, thus ensuring the physical meaning and logical correctness of subsequent ratio calculations.

[0054] (2) Divide the number of input / output operations per second collected in real time by the associated maximum allowable value to obtain the first ratio.

[0055] After obtaining the real-time value of the number of input / output operations per second and the corresponding maximum allowable value determined in step (1), the first ratio is obtained by calculating the real-time value ÷ the maximum allowable value. This ratio reflects the proportion of the current IOPS load to its upper limit. The larger the value, the heavier the load on the hard drive in terms of operation frequency. When the ratio is close to 1, it indicates that the IOPS is close to the upper limit under the current hardware and demand constraints.

[0056] (3) Divide the real-time hard disk data transfer rate by the associated maximum allowable value to obtain the second ratio.

[0057] For hard drive data transfer rates, the second ratio is obtained by directly dividing the real-time value by the corresponding maximum allowable value. This ratio intuitively reflects the proportion of data transfer load to its upper limit and is a key indicator for measuring the overall load of the hard drive. Because it is consistent with the current maximum allowable performance parameter, the calculation process does not require additional conversion, and the physical meaning of the result is also the most direct.

[0058] (4) Divide the real-time acquisition seek delay by the associated maximum allowable value to obtain the third ratio.

[0059] The calculation logic for the seek latency ratio is the same as the previous two. The real-time value (e.g., 16 milliseconds at a certain acquisition time) is divided by the maximum allowable value determined in step (1) (e.g., 20 milliseconds) to obtain the third ratio (e.g., 16 ÷ 20 = 0.8). Unlike the previous two ratios, the larger the seek latency ratio, the lower the current head positioning efficiency and the heavier the load (high latency means frequent head movement or long distance). This ratio supplements the load assessment dimension from the perspective of hard drive mechanical operation efficiency.

[0060] Through the above four steps, the real-time performance parameters that were originally of different dimensions are uniformly transformed into ratios between 0 and 1, which not only preserves the characterization characteristics of each parameter on the load, but also achieves data standardization, thus providing a foundation for subsequent operations.

[0061] S103. Calculate the weighted sum of each ratio to obtain the dynamic load score of the hard disk.

[0062] It's important to note that after obtaining the ratios of each real-time read / write performance parameter to its corresponding maximum allowable value (i.e., the first ratio, the second ratio, and the third ratio), a weighted sum of these ratios is needed to obtain the hard drive's dynamic load score. This integrates the load percentages from multiple dimensions (IOPS percentage, transfer rate percentage, and seek latency percentage) into a comprehensive indicator, enabling a quantitative assessment of the hard drive's overall load status. Compared to single parameter ratios, the dynamic load score more comprehensively reflects the actual workload of the hard drive. For example, in a scenario where the IOPS ratio is low but the transfer rate ratio is high, a single parameter might misjudge it as low load, while the weighted sum can combine the weights of both to determine a medium-to-high load, providing a more accurate basis for subsequent fan speed adjustments.

[0063] It should be noted that the key to the weighted sum calculation lies in the weights corresponding to each ratio. These weights need to be predetermined through offline calibration. Specifically, the calculation of the weights for the weighted sum of the ratios includes:

[0064] (1) Under various preset load scenarios, apply load to the hard disk to collect historical data of multiple real-time read and write performance parameters, and label the historical data with the corresponding benchmark load score.

[0065] It's important to note that the foundation of weight calibration lies in constructing a sufficiently comprehensive and accurate dataset. In a laboratory environment, the system applies gradient loads from low to high to the target model's hard drive under various preset load scenarios, such as idle states, typical office application loads, and continuous high-intensity database operations or large-scale data backups. During this process, the system synchronously and frequently collects historical data on multiple real-time read / write performance parameters (here, real-time read / write performance parameters refer to the aforementioned hard drive data transfer rate, input / output operations per second, and seek latency), forming a large number of data samples. Furthermore, a corresponding baseline load score needs to be labeled for each set of collected performance parameter data. This baseline score is a true value representing the actual comprehensive load level at that moment. It can be obtained by conversion using a high-precision external power monitoring device, or assigned a value by domain experts based on comprehensive information such as the load scenario and system resource utilization.

[0066] (2) Using regression analysis, fit the nonlinear relationship between each real-time read and write performance parameter and the benchmark load score to determine the degree of influence of each real-time read and write performance parameter on the load.

[0067] Specifically, for the historical data collected in step (1), regression analysis (such as nonlinear regression, multinomial regression, etc.) needs to be performed separately for each real-time read and write performance parameter. The historical data value of the number of input and output operations per second is used as the horizontal axis, and the benchmark load score in the corresponding scenario is used as the vertical axis to fit a nonlinear curve of IOPS and benchmark score. Similarly, the historical data values ​​of hard disk data transfer rate and seek latency are used as the horizontal axis, and the benchmark load score is used as the vertical axis to fit two other nonlinear curves.

[0068] To quantify the impact of each parameter, the system calculates the average absolute slope of each fitted curve within a typical working range (e.g., a range where the baseline load score is from 0.2 to 0.8). This average absolute slope represents the initial impact of the corresponding real-time read / write performance parameter. Specifically, the larger the slope, the more drastic the change in overall load caused by a unit change in that performance parameter, and the higher its weight.

[0069] For example, if the average absolute slope of the curve corresponding to the hard disk data transfer rate is larger, it means that the change in the baseline load score is more significant for every unit change in the transfer rate, that is, the parameter has a higher initial impact on the load; conversely, if the average absolute slope of the curve corresponding to the seek latency is smaller, it means that its initial impact is relatively lower.

[0070] (3) Normalize the degree of influence and make corrections based on the sum of the normalized degree of influence.

[0071] After obtaining the influence of each parameter through step (2), these influence levels need to be normalized to ensure that the sum of the final determined weights is 1. This can avoid the dynamic load score from exceeding a reasonable range (e.g., exceeding 1.0 or falling below 0) due to the total weight being greater than or less than 1. In specific operation, first calculate the sum of the initial influence of all real-time read and write performance parameters. Then, divide the initial influence of each parameter by the sum to obtain its initial weight. If there are extreme values ​​in the initial weight (e.g., the weight of a certain parameter is close to 1, and the others are close to 0), it can be finely adjusted according to the actual scenario requirements to finally obtain a set of weights with a total sum of 1. Specifically, the system optimizes and adjusts the initial weights according to the preset correction rules to prevent the load assessment dimension from becoming unbalanced due to the weight of a certain parameter being too high or too low. The correction rules may include setting a lower limit (e.g., 0.05) for the weight of each parameter. When the initial weight is lower than this lower limit, it is adjusted to this lower limit value, and the weights of other parameters are reduced proportionally, and all weights are ensured to be 1 again.

[0072] (4) The modified degree of influence is used as the weight.

[0073] The set of weights determined through the aforementioned offline calibration process (e.g., hard disk data transfer rate weight 0.6, input / output operations per second weight 0.3, seek latency weight 0.1) will be preset into the fan speed control system. During real-time operation, when calculating the dynamic load score, the system multiplies the collected first ratio, second ratio, and third ratio by their respective preset weights, and then sums all the products to obtain the dynamic load score, which comprehensively reflects the hard disk load status. The specific calculation formula is as follows: Dynamic Load Score = (First Ratio × Corresponding Weight) + (Second Ratio × Corresponding Weight) + (Third Ratio × Corresponding Weight).

[0074] S104. Calculate the speed control value based on the nonlinear relationship between the dynamic load score and the fan speed, wherein the nonlinear relationship is a piecewise linear function, and the dynamic load score is determined from multiple piecewise linear functions to identify the target piecewise linear function.

[0075] It should be noted that after obtaining the dynamic load score of the hard drive, the fan speed control value needs to be calculated based on the non-linear relationship between the score and the fan speed. In this application, this non-linear relationship is specifically implemented by a piecewise linear function, which divides the continuous range of the dynamic load score into three load intervals: low, medium, and high, by setting a first segment threshold and a second segment threshold. Specifically, in the low-load range (dynamic load score ≤ first segment threshold), the hard drive generates little heat and has low cooling requirements. The piecewise function outputs a constant minimum safe rotational speed within this range, maximizing energy saving and noise reduction while meeting basic cooling needs. In the medium-load range (first segment threshold < dynamic load score ≤ second segment threshold), hard drive heat generation increases approximately linearly with the load. The piecewise function employs a linearly increasing rotational speed strategy within this range, ensuring that cooling capacity changes smoothly and synchronously with the heat load, avoiding both response lag and sudden changes in rotational speed. In the high-load range (dynamic load score > second segment threshold), the hard drive generates enormous heat and has the highest cooling requirements. The piecewise function outputs a constant maximum allowable rotational speed within this range, providing maximum airflow and prioritizing temperature safety for the hard drive under extreme loads. It should also be noted that in practical applications, multiple piecewise linear functions (i.e., with different thresholds and scaling factors) may be preset for different hard drive models and usage scenarios. The system will automatically match the corresponding target piecewise linear function based on the current dynamic load score range, ensuring that the speed adjustment strategy is adapted to the specific hardware characteristics and working scenario.

[0076] Specifically, the piecewise linear function includes a first piecewise threshold and a second piecewise threshold. Before calculating the speed control value based on the nonlinear relationship between the dynamic load score and the fan speed, the following steps are included:

[0077] (1) Under the preset test environment conditions, apply a continuous gradient load to the hard disk from no load to full load.

[0078] First, a standardized testing environment is established. This environment requires maintaining stable external variables, such as a constant ambient temperature of 25 degrees Celsius, and ensuring that the chassis airflow is not interfered with by other devices' heat dissipation. This guarantees the accuracy of the collected data, allowing it to purely reflect the intrinsic relationship between hard drive load and its own temperature. Next, a professional load generation tool is used to apply a continuous gradient load from idle to full load to the target hard drive. The idle state represents the basic operating state of the hard drive without any read or write operations. The load is applied gradually according to a preset gradient step size; for example, the read / write request volume can be increased by 10% each time. This process continues until the hard drive's performance parameters reach its current maximum allowable value, which is considered to be entering the full load state. This gradient loading method systematically covers all possible load levels from the lowest to the highest, thus providing a complete and continuous data sample for subsequent synchronous collection of dynamic load scores and hard drive steady-state operating temperatures.

[0079] (2) Under each load level, the dynamic load score and the corresponding hard disk steady-state operating temperature are collected synchronously. The hard disk steady-state operating temperature is the temperature value that fluctuates within multiple consecutive collection cycles without exceeding a preset value.

[0080] Data collection should only commence after the hard drive reaches thermal equilibrium at each load level. Thermal equilibrium refers to a balance between the hard drive's heat generation and heat dissipation efficiency, where its operating temperature no longer exhibits a continuous upward or downward trend, but fluctuates only within a very small range. To objectively determine this state and collect the steady-state operating temperature, the system continuously monitors the hard drive temperature after applying each load gradient. For example, thermal equilibrium is considered reached when the rate of temperature change is below 0.5 degrees Celsius per minute for two consecutive minutes. Subsequently, the system continuously collects at least 10 temperature samples at a fixed sampling period (recommended period is 10 seconds). The steady-state operating temperature of the hard drive is the arithmetic mean of all sample values ​​in this set of continuous samples where the difference between the maximum and minimum values ​​does not exceed a preset stability threshold (recommended value is 1 degree Celsius).

[0081] This method ensures that the obtained temperature data accurately reflects the stable heat generation under the load, effectively avoiding threshold misjudgments caused by instantaneous thermal disturbances or measurement noise. For example, after a certain load level reaches thermal equilibrium, if 10 consecutively collected temperature values ​​are 42.1℃, 42.3℃, 41.9℃, 42.2℃, 42.0℃, 42.4℃, 42.1℃, 42.2℃, 41.8℃, and 42.3℃, and the maximum difference between them is 0.6 degrees Celsius, which is below the stability threshold of 1 degree Celsius, then their arithmetic mean can be calculated and identified as the steady-state operating temperature under that load.

[0082] (3) Using the upper limit of the safe operating temperature of the hard disk as a constraint, based on the collected dynamic load score and the corresponding data of the hard disk steady-state operating temperature, the dynamic load score critical point is selected; when the dynamic load score is lower than the critical point, the hard disk steady-state operating temperature is lower than the upper limit of the safe operating temperature and the fan speed remains unchanged; when the dynamic load score is higher than the critical point, the fan speed is increased to maintain the hard disk steady-state operating temperature from not exceeding the upper limit of the safe operating temperature.

[0083] Based on the collected dynamic load score and steady-state operating temperature data, and combined with the hard drive's safe operating temperature limit (usually specified by the manufacturer, such as 60℃), a critical point is selected. Specifically, all data points are sorted from low to high according to their dynamic load scores. Starting from the data point with the lowest score, each data point is traversed sequentially. If the steady-state operating temperature corresponding to the current data point and all previous data points is lower than the safe operating temperature limit, and the system records show that the fan can maintain the minimum speed to meet cooling requirements at these points, then the current state belongs to the range where no speed adjustment is needed. When a data point is traversed and its corresponding steady-state operating temperature reaches or exceeds the safe operating temperature limit for the first time, this data point marks a critical state. At this time, if the fan speed remains unchanged, the hard drive temperature will exceed the safe range. The dynamic load score corresponding to the data point that first triggers the temperature over-limit is determined as a dynamic load score critical point. This critical point is the boundary between the two states where no speed increase is needed and where a speed increase is necessary. In addition, after completing the first traversal, we continue to check all subsequent data points with higher scores to confirm that their steady-state temperatures have exceeded the limit before the fan speed is increased, so as to ensure that no lower score critical points are missed.

[0084] (4) From the selected dynamic load score critical points, determine the first segment threshold and the second segment threshold; wherein, the critical point with the smallest difference between the corresponding hard disk steady-state operating temperature and the preset low temperature target value is taken as the first segment threshold; and the critical point with the smallest difference between the corresponding hard disk steady-state operating temperature and the upper limit of the hard disk safe operating temperature is taken as the second segment threshold.

[0085] It should be noted that in step (3), through the gradient load experiment, the system will select multiple dynamic load score critical points. Each critical point represents the load state at which the fan speed needs to be increased. In order to construct a piecewise linear function, two key points need to be selected from these critical points as the first segment threshold (the boundary between low load and medium load) and the second segment threshold (the boundary between medium load and high load). Specifically, among all the selected critical points, the absolute difference between the hard drive steady-state operating temperature and the preset low temperature target value (e.g., ambient temperature +5℃ or 30℃) corresponding to each critical point is calculated; the critical point with the smallest absolute difference is selected, and its corresponding dynamic load score is used as the first segment threshold. Below this threshold, the hard drive generates little heat, and the fan can meet the heat dissipation requirements by maintaining the minimum safe speed.

[0086] Among all the selected critical points, the absolute difference between the hard drive's steady-state operating temperature and its safe operating temperature limit (e.g., 60°C) is calculated for each critical point. The critical point with the smallest absolute difference is selected, and its corresponding dynamic load score is used as the second segment threshold. Above this threshold, the hard drive's heat generation is close to the safe limit, and the fan needs to run at its maximum allowable speed to enhance heat dissipation. Through the above operations, two key thresholds are clearly selected from multiple critical points, dividing the dynamic load score into three load ranges: low, medium, and high, providing a clear and reasonable basis for the piecewise linear function.

[0087] Based on the first and second segmented thresholds determined above, the speed control rule for the piecewise linear function is as follows:

[0088] (1) When the dynamic load score is less than or equal to the first segment threshold, the speed control value corresponds to the minimum safe speed of the fan.

[0089] At this time, the hard drive is under low load, generates little heat, and the steady-state temperature is far below the safe upper limit. The fan only needs to maintain the minimum safe speed to meet the heat dissipation requirements. This speed must ensure that the temperature of the hard drive does not exceed the preset low temperature target value during the longest continuous low load operation, while minimizing noise and energy consumption.

[0090] (2) When the dynamic load score is greater than the first segment threshold and less than or equal to the second segment threshold, the speed control value increases linearly between the minimum safe speed and the maximum allowable speed.

[0091] At this point, the hard drive is under medium load, and its heat generation increases with the increase in load. Dynamic heat dissipation balance needs to be achieved by linearly increasing the rotational speed. For example, if the first threshold is 0.3, the second threshold is 0.8, the minimum rotational speed is 800 RPM, and the maximum rotational speed is 3000 RPM, then when the score increases from 0.3 to 0.8, the rotational speed linearly increases from 800 RPM to 3000 RPM, ensuring that the heat dissipation capacity matches the increase in load and preventing a sudden rise in temperature.

[0092] Specifically, the speed control value increases linearly proportionally between the minimum safe speed and the maximum permissible speed, including:

[0093] (i) Calculate the difference between the first segment threshold and the second segment threshold, and use the difference as the interval division benchmark.

[0094] For example, if the first threshold is 0.3 and the second threshold is 0.8, the difference is 0.5. This value represents the total span of the dynamic load score in the medium load range and is the basic parameter for subsequent ratio calculations.

[0095] (ii) Collect multiple sets of dynamic load scores and corresponding optimal fan speeds through gradient load experiments. The optimal fan speed is the minimum speed that keeps the hard drive temperature stable within a safe range.

[0096] Within the medium load range, multiple load points are set in smaller increments (e.g., 0.05), each point corresponding to a dynamic load score. Through experiments, the optimal fan speed corresponding to each score is found, which is the minimum speed that keeps the hard drive temperature stable within the safe range (e.g., at a score of 0.4, 1200 RPM can keep the temperature stable at 40℃, and reducing the speed will cause the temperature to exceed the safe value, so 1200 RPM is the optimal speed). These matching data reflect the precise correspondence between load and speed.

[0097] (iii) Based on the matching data, a linear relationship between the dynamic load score and the optimal fan speed within the interval between the first segment threshold and the second segment threshold is determined by linear regression, and the slope of the linear relationship is determined as the ratio.

[0098] Specifically, based on the matching data of multiple sets of dynamic load scores and corresponding optimal fan speeds collected in step (ii), a linear regression method is used to fit a straight line. This straight line describes the linear relationship between the dynamic load score and the optimal fan speed within the medium load range. The slope of this line is the proportionality, representing the amount by which the fan speed should increase when the dynamic load score increases by one unit. For example, if the equation of the straight line obtained through linear regression is speed = slope × dynamic load score + intercept, then this slope is used as the proportionality parameter. This proportionality parameter ensures that within the medium load range, the fan speed can be adjusted linearly and smoothly according to the changes in the dynamic load score, which not only conforms to the actual measured data but also achieves a balance between heat dissipation effect and energy consumption.

[0099] (3) When the dynamic load score is greater than the second segment threshold, the speed control value corresponds to the maximum allowable speed of the fan.

[0100] At this time, the hard drive is under high load, and the heat generation is close to or reaches the safe limit. The fan needs to run at the highest allowable speed to maximize the heat dissipation capacity and ensure that the hard drive temperature does not exceed the safe operating temperature limit, so as to prioritize hardware stability.

[0101] In addition, the calculation of the maximum permissible speed and the minimum safe speed is explained. Specifically, a fixed minimum safe speed and maximum permissible speed can be preset based on the hard drive model, chassis cooling specifications, and historical operating data. The minimum safe speed should ensure that the hard drive maintains basic cooling requirements when idle or under very low load, which is usually 20%-30% of the fan's rated speed. The maximum permissible speed is based on ensuring that the hard drive temperature does not exceed the safe operating limit under full load, which is usually 80%-100% of the fan's rated speed.

[0102] It should also be noted that a heat conduction equation can be established based on the physical layout of the hard drives and fans within the chassis and the characteristics of the heat dissipation airflow. Combined with fan performance curves, the minimum safe speed and maximum allowable speed for each fan can be calculated. Specifically, first, a heat generation model for the hard drive group is established, calculating the heat generation of each group of hard drives under different loads based on the power consumption characteristics of the hard drives. Second, a heat dissipation model for the fan group is established, calculating the heat dissipation efficiency of each fan for the hard drive group based on the fan airflow-speed characteristic curves and their position distribution. Finally, the optimal speed range is solved to ensure that the temperature of all hard drives remains within a safe range. The minimum safe speed (the minimum value to meet basic heat dissipation requirements) and maximum allowable speed (the maximum value to meet extreme heat dissipation requirements) required for each fan are solved using a heat balance equation, achieving the optimal balance between heat dissipation efficiency and energy consumption.

[0103] S105. Control the fan to rotate according to the speed control value.

[0104] After calculating the fan speed control value, the control value needs to be sent to the fan controller (such as the PWM controller on the motherboard or an independent heat dissipation control module). The controller adjusts the actual rotation speed of the fan according to the control value. For example, if the speed control value corresponds to 1500 rpm, the controller will output the corresponding PWM signal (pulse width modulation signal) to drive the fan motor to run at that speed, thereby ensuring that the fan's heat dissipation capacity matches the real-time heat generation of the hard drive.

[0105] It should be noted that calculating the rotational speed using only a fixed piecewise linear function may lead to discrepancies between the cooling effect corresponding to the speed control value and the actual cooling requirements of the hard drive due to factors such as hard drive aging, changes in ambient temperature (e.g., increased overall temperature inside the chassis during summer), and sudden load fluctuations. For example, a fan operating at the control value may fail to reduce the hard drive temperature to the expected range, or excessive cooling may result in wasted energy. Therefore, this method also incorporates an adaptive correction step, dynamically adjusting the speed control parameters through real-time temperature feedback to further improve control accuracy. Furthermore, for complex scenarios with multiple hard drives and fans (e.g., multiple hard drive arrays within a server chassis), a grouped collaborative speed control mechanism is designed to ensure optimal overall cooling efficiency.

[0106] Specifically, after controlling the fan to rotate according to the speed control value, the following steps are taken:

[0107] (1) Real-time acquisition of the actual temperature of the hard drive.

[0108] During the operation of the fan at the speed control value, the actual operating temperature of the hard drive needs to be collected in real time at a fixed interval (e.g., 2 seconds / time) using the temperature sensor built into the hard drive or an external temperature probe. The collection frequency needs to be higher than the adjustment cycle of load and speed to ensure that the temperature change trend can be captured in time and provide real-time data support for deviation judgment.

[0109] (2) Calculate the deviation between the actual temperature and the temperature predicted based on the dynamic load score.

[0110] Based on the current dynamic load score of the hard drive, and combined with a preset load-temperature prediction model (built through offline experiments, storing the mapping relationship between different dynamic load scores and corresponding stable temperatures), the predicted temperature at that score is calculated. Then, the temperature deviation is obtained by subtracting the absolute value of the predicted temperature from the actual temperature. For example, if the predicted temperature corresponding to a current dynamic load score of 0.6 is 45℃, and the actual collected temperature is 48℃, then the deviation is 3℃. By comparing the predicted value with the actual value, it can be determined whether the current speed control value can meet the heat dissipation requirements.

[0111] (3) When the absolute value of the deviation exceeds the first preset threshold, the value of the proportional parameter of the piecewise linear function is adjusted by a preset step size.

[0112] A first preset threshold is set (e.g., ±2℃, which can be adjusted according to the hard drive's sensitivity to temperature fluctuations). If the absolute value of the temperature deviation exceeds this threshold, it indicates that the proportional parameter of the current piecewise linear function (i.e., the correlation strength between the score in the medium load range and the rotational speed) is no longer applicable and dynamic correction is required. If the actual temperature is higher than the predicted temperature (positive deviation), it indicates that the current rotational speed is too low and the heat dissipation is insufficient. The proportional parameter needs to be increased by a preset step size (e.g., increasing by 0.1 each time) to make the rotational speed control value corresponding to the same dynamic load score higher. If the actual temperature is lower than the predicted temperature (negative deviation), it indicates that the current rotational speed is too high and the heat dissipation is excessive. The proportional parameter needs to be decreased by a preset step size (e.g., decreasing by 0.1 each time) to make the rotational speed control value corresponding to the same score lower. The preset step size needs to balance adjustment accuracy and stability to avoid frequent fluctuations in rotational speed due to excessively large step sizes.

[0113] (4) Using the adjusted proportional parameters, recalculate and correct the fan speed control value.

[0114] After adjusting the proportional parameters, the piecewise linear function in S104 needs to be re-substituted into the current dynamic load score to recalculate the speed control value. For example, if the original proportional parameter was 4400 rpm / score, and the adjusted parameter is 4840 rpm / score, with a current score of 0.6, the corrected speed control value will be increased from the original calculated value (e.g., 2000 rpm) to a higher value (e.g., 2200 rpm). The corrected control value is then sent to the fan controller to update the fan speed, ensuring that the cooling effect always matches the actual needs during long-term operation.

[0115] Furthermore, when there are multiple hard drives and multiple fans in the system, the hard drives and fans are grouped according to their physical locations, and a group correspondence between fans and hard drives is established. Each group includes multiple fans and multiple hard drives, and the fans and hard drives in different groups do not overlap. For each group, the rotation speed control value of the current group is calculated based on the average real-time read and write performance parameters of each group of hard drives. The rotation speed of the fans in the adjacent groups is adjusted according to the relationship between the rotation speed control value of the current group and the preset threshold.

[0116] Specifically, when a system contains multiple hard drives and multiple fans (such as multiple hard drive cages physically divided within a server chassis), if independent speed control is used for each hard drive, the heat dissipation of adjacent hard drives may affect each other (e.g., high load on one group of hard drives causing a passive temperature increase in adjacent groups), leading to localized overheating or overall low heat dissipation efficiency. Therefore, it is necessary to group hard drives and fans according to their physical location. Specifically, first, establish a grouping relationship based on the physical layout of hard drives and fans (e.g., hard drive cage partitions within the chassis, and the heat dissipation coverage of fan installation locations). Group multiple hard drives located in the same heat dissipation area and covered by the same fan into one group, and assign corresponding multiple fans to this group (e.g., 8 hard drives in a hard drive cage + 2 front fans in the corresponding location form a group), ensuring that fans and hard drives in different groups do not overlap, avoiding overlapping or missed heat dissipation responsibilities.

[0117] Secondly, for each group, the average real-time read and write performance parameters of all hard drives in the group (such as average IOPS, average transfer rate, and average seek latency) are calculated. Then, the average dynamic load score of the group is calculated according to the methods in S102-S104. Finally, the speed control value of the current group is obtained. In this way, it can be ensured that the speed adjustment of the group can reflect the overall load level of the hard drives in the group, rather than the local load of a single hard drive.

[0118] Finally, based on the relationship between the current group's fan speed control value and the preset threshold, the fan speeds of adjacent groups are adjusted to achieve chassis-level collaborative heat dissipation optimization. In practice, firstly, a grouping correspondence is established based on the physical layout of hard drives and fans within the server chassis. Multiple hard drives in the same heat dissipation area with similar heat dissipation characteristics are grouped into a hard drive group, and multiple fans jointly responsible for heat dissipation in that area are grouped into a fan group, ensuring that the coverage of each group does not overlap and that the division of labor is clear. For each hard drive group, a comprehensive load score is calculated based on the real-time read / write performance parameters of each hard drive within the group, and then the corresponding basic fan speed control value is obtained through a piecewise linear function. When a fan group contains multiple fans, the system calculates the optimal fan speed allocation scheme within the group based on the total heat generation of the hard drive group and the performance curves of each fan, aiming to meet the overall heat dissipation requirements while minimizing the total power consumption of the fan group. This typically tends to allow all fans to work collaboratively at medium speeds, rather than a single fan operating at high speed, thereby achieving optimal energy efficiency.

[0119] Building upon group control, the system further considers the mutual influence of heat dissipation between groups, establishing a heat diffusion model to quantitatively evaluate the heat diffusion effect of high-load hard drive groups on adjacent areas and the auxiliary cooling effect of airflow generated by adjacent fan groups on the current group. When the speed control value of a certain hard drive group exceeds the preset high-load threshold (e.g., 2800 RPM), it indicates that the group is in a high-load, high-heat state, and its heat dissipation demand may exceed the independent cooling capacity of its fans. At this time, the system calculates the additional cooling efficiency required to offset the heat diffusion effect and fully utilize the cooling potential of adjacent fan groups based on the heat diffusion model, and converts this efficiency requirement into an increase in the speed control value of adjacent fan groups (e.g., an increase based on a certain proportion of the original calculated value). This adjustment process, through dynamic iterative optimization, ensures that the temperature of all hard drives is maintained within a safe range, while pursuing the minimization of the total power consumption of the entire system's fans, thereby achieving optimal energy consumption control while ensuring heat dissipation effect.

[0120] The method provided in this embodiment is based on real-time hard drive read / write performance parameters. It calculates the ratio of these parameters to the maximum allowable performance parameters determined in real-time by the hard drive space layout and the total requested computational load. This ratio is then weighted using regression fitting calibration to obtain a dynamic load score, ensuring that the load assessment accurately reflects the actual workload of the hard drive. Simultaneously, it determines dual-segment thresholds based on hard drive safety temperature constraints. A piecewise linear function is used to achieve differentiated control: low speed under low load, linear speed adjustment under medium load, and high speed under high load. This avoids the ineffective energy consumption of traditional fixed-speed strategies and solves the response lag problem of single-temperature-triggered strategies, allowing for precise matching of fan speed and hard drive heat generation, balancing heat dissipation effectiveness and energy-saving requirements. Furthermore, adjusting the piecewise function ratio parameters through real-time temperature feedback dynamically corrects speed deviations caused by hard drive aging and environmental temperature changes, ensuring stable heat dissipation during long-term operation. The grouped collaborative speed adjustment mechanism is designed for multi-hard drive, multi-fan scenarios. It groups fans by physical location and calculates the speed based on the average parameters within each group, while simultaneously adjusting fans in adjacent groups to avoid localized overheating and overall low heat dissipation efficiency, adapting to the needs of multi-hard drive applications such as servers.

[0121] Example 2

[0122] Corresponding to the aforementioned embodiment of a fan speed control method based on hard disk read / write performance monitoring, this application also provides an embodiment of a fan speed control device based on hard disk read / write performance monitoring.

[0123] Figure 2 This is a schematic diagram of Embodiment 2 of the fan speed control device based on hard disk read / write performance monitoring provided in this application. Please refer to... Figure 2 The apparatus provided in this embodiment includes an acquisition module 210, a calculation module 220, and a processing module 230;

[0124] The acquisition module 210 is used to acquire multiple real-time read and write performance parameters of the hard disk;

[0125] The calculation module 220 is used to calculate the ratio of the real-time read and write performance parameters of each hard disk to the current maximum allowable performance parameters of the hard disk, wherein the current maximum allowable performance parameters of the hard disk are calculated in real time by the space layout of the hard disk and the total amount of data read and write requests.

[0126] The calculation module 220 is also used to calculate the weighted sum of each ratio to obtain the dynamic load score of the hard disk;

[0127] The calculation module 220 is further configured to calculate a speed control value based on the nonlinear relationship between the dynamic load score and the fan speed, wherein the nonlinear relationship is a piecewise linear function, and the dynamic load score is determined from multiple piecewise linear functions to identify a target piecewise linear function.

[0128] The processing module 230 is used to control the fan rotation according to the speed control value.

[0129] The apparatus of this embodiment can be used to perform... Figure 1 The steps of the method embodiment shown are similar in principle and process, and will not be repeated here.

[0130] The specific implementation process of the functions and roles of each unit 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.

[0131] 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 units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. 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.

[0132] The above description is merely a preferred embodiment of this application and is not intended to limit this application. 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 fan speed regulation method based on hard disk read-write performance monitoring, characterized in that, The method comprises: acquiring a plurality of real-time read-write performance parameters of the hard disk; the real-time read-write performance parameters at least include the number of input-output operations per second, the hard disk data transmission rate and the seek delay; calculating the ratio of each real-time read-write performance parameter of the hard disk to the current maximum allowable performance parameter of the hard disk, wherein the current maximum allowable performance parameter of the hard disk is calculated in real time based on the spatial layout of the hard disk and the total operation amount of data read-write requests; the calculation of the ratio of each real-time read-write performance parameter of the hard disk to the current maximum allowable performance parameter of the hard disk comprises: determining the maximum allowable value associated with the number of input-output operations per second, the hard disk data transmission rate and the seek delay respectively based on the determined current maximum allowable performance parameter; dividing the real-time collected number of input-output operations per second by the associated maximum allowable value to obtain a first ratio; dividing the real-time collected hard disk data transmission rate by the associated maximum allowable value to obtain a second ratio; dividing the real-time collected seek delay by the associated maximum allowable value to obtain a third ratio; calculating the weighted sum of each ratio to obtain a dynamic load score of the hard disk; calculating a rotation speed control value according to the nonlinear relationship between the dynamic load score and the fan rotation speed, wherein the nonlinear relationship is a piecewise linear function, and the dynamic load score determines a target piecewise linear function from a plurality of piecewise linear functions; controlling the rotation of the fan according to the rotation speed control value; The piecewise linear function comprises a first segment threshold and a second segment threshold, and before calculating the rotation speed control value according to the nonlinear relationship between the dynamic load score and the fan rotation speed, it comprises: applying a continuous gradient load covering from no load to full load to the hard disk under a preset test environment condition; synchronously collecting the dynamic load score and the corresponding hard disk steady-state working temperature at each load level, wherein the hard disk steady-state working temperature is a temperature value that fluctuates by no more than a preset value within a plurality of consecutive collection periods; With the upper limit of the safe working temperature of the hard disk as a constraint condition, based on the collected corresponding data of the dynamic load score and the hard disk steady-state working temperature, a dynamic load score critical point is screened; when the dynamic load score is lower than the critical point, the hard disk steady-state working temperature is lower than the upper limit of the safe working temperature and the fan rotation speed is unchanged; when the dynamic load score is higher than the critical point, the fan rotation speed is increased to maintain the hard disk steady-state working temperature not exceeding the upper limit of the safe working temperature; From the screened dynamic load score critical point, the first segment threshold and the second segment threshold are determined; wherein the critical point corresponding to the smallest difference between the hard disk steady-state working temperature and the preset low-temperature target value is taken as the first segment threshold; the critical point corresponding to the smallest difference between the hard disk steady-state working temperature and the upper limit of the hard disk safe working temperature is taken as the second segment threshold.

2. The method of claim 1, wherein, Before calculating the ratio of each real-time read-write performance parameter of the hard disk to the current maximum allowable performance parameter of the hard disk, it comprises: determining the parallel read-write bandwidth of each storage area based on the physical spatial layout information of the hard disk within a preset time window; statistically calculating the total operation amount of data read-write requests within the preset time window; The parallel read-write bandwidth is compared with the total operation amount of the data read-write request, and the smaller one is taken as a current maximum allowed performance parameter of the hard disk at the current time.

3. The method of claim 1, wherein, The weight calculation of the weighted sum of the ratio includes: In a plurality of preset load scenarios, a plurality of historical data of real-time read-write performance parameters are collected by applying a load to the hard disk, and a corresponding benchmark load score is labeled for the historical data; By using regression analysis, a nonlinear relationship between each real-time read-write performance parameter and the benchmark load score is fitted to determine the influence degree of each real-time read-write performance parameter on the load; The influence degree is normalized, and the sum of the normalized influence degrees is corrected; The corrected influence degree is used as the weight.

4. The method of claim 1, wherein, The segmented linear function includes: When the dynamic load score is less than or equal to a first segmented threshold, the speed control value corresponds to the lowest safe speed of the fan; When the dynamic load score is greater than the first segmented threshold and less than or equal to a second segmented threshold, the speed control value is proportionally and linearly increased between the lowest safe speed and the highest allowed speed; When the dynamic load score is greater than the second segmented threshold, the speed control value corresponds to the highest allowed speed of the fan.

5. The method of claim 4, wherein, The proportionally and linearly increased speed control value between the lowest safe speed and the highest allowed speed includes: The difference between the first segmented threshold and the second segmented threshold is calculated, and the difference is used as an interval division reference; A plurality of matching data of dynamic load scores and corresponding optimal fan speeds are collected through gradient load experiments, and the optimal fan speed is the minimum speed that stabilizes the hard disk temperature within a safe interval; Based on the matching data, a linear relationship between the dynamic load score and the optimal fan speed in the interval between the first segmented threshold and the second segmented threshold is determined by linear regression, and the slope of the linear relationship is determined as the proportion.

6. The method of claim 1, wherein, After the fan is controlled to rotate according to the speed control value, it includes: The actual temperature of the hard disk is collected in real time; The deviation between the actual temperature and the predicted temperature based on the dynamic load score is calculated; When the absolute value of the deviation exceeds a first preset threshold, the value of the proportion parameter of the segmented linear function is adjusted by a preset step; The speed control value of the fan is recalculated and corrected using the adjusted proportion parameter.

7. The method of claim 1, wherein, According to the physical positions of the hard disks and the fans, a grouping relationship between the fans and the hard disks is established, each group including a plurality of fans and a plurality of hard disks, and the fans and the hard disks in different groups do not overlap; for each group, the speed control value of the current group is calculated based on the mean value of the real-time read-write performance parameters of each hard disk in the group; the speed of the fan in a group that is less than a preset threshold from the current group is adjusted according to the relationship between the speed control value of the current group and the preset threshold.

8. A fan speed regulation device based on hard disk read-write performance monitoring, characterized in that, The device includes an acquisition module, a calculation module, and a processing module; The acquisition module is configured to acquire a plurality of real-time read-write performance parameters of a hard disk; the real-time read-write performance parameters include at least the number of input-output operations per second, the hard disk data transmission rate, and the seek delay. The computing module is configured to calculate a ratio of a real-time read-write performance parameter of each hard disk to a current maximum allowable performance parameter of the hard disk, wherein the current maximum allowable performance parameter of the hard disk is calculated in real time based on a space layout of the hard disk and a total operation amount of data read-write requests; and the computing of the ratio of the real-time read-write performance parameter of each hard disk to the current maximum allowable performance parameter of the hard disk includes: determining a maximum allowable value associated with an input / output operation per second, a hard disk data transmission rate and a seek delay, respectively, based on the determined current maximum allowable performance parameter; dividing a real-time collected input / output operation per second by the associated maximum allowable value to obtain a first ratio; dividing a real-time collected hard disk data transmission rate by the associated maximum allowable value to obtain a second ratio; and dividing a real-time collected seek delay by the associated maximum allowable value to obtain a third ratio. The computing module is further configured to calculate a weighted sum of the ratios to obtain a dynamic load score of the hard disk. The computing module is further configured to calculate a rotation speed control value based on a nonlinear relationship between the dynamic load score and a fan rotation speed, wherein the nonlinear relationship is a piecewise linear function, and the dynamic load score determines a target piecewise linear function from a plurality of piecewise linear functions. The processing module is configured to control rotation of the fan based on the rotation speed control value. The piecewise linear function includes a first segment threshold and a second segment threshold, and before the calculating of the rotation speed control value based on the nonlinear relationship between the dynamic load score and the fan rotation speed, the method includes: applying a continuous gradient load covering from an empty load to a full load to the hard disk under preset test environment conditions; synchronously collecting a dynamic load score and a corresponding hard disk steady-state operating temperature at each load level, wherein the hard disk steady-state operating temperature is a temperature value that fluctuates by no more than a preset value within a plurality of consecutive collection periods; filtering a dynamic load score critical point based on the collected corresponding data of the dynamic load score and the hard disk steady-state operating temperature, with a hard disk safe operating temperature upper limit as a constraint condition; when the dynamic load score is lower than the critical point, the hard disk steady-state operating temperature is lower than the safe operating temperature upper limit and the fan rotation speed is unchanged; and when the dynamic load score is higher than the critical point, the fan rotation speed is increased to maintain the hard disk steady-state operating temperature not to exceed the safe operating temperature upper limit; determining the first segment threshold and the second segment threshold from the filtered dynamic load score critical points; wherein a critical point corresponding to a hard disk steady-state operating temperature that has the smallest difference from a preset low-temperature target value is taken as the first segment threshold; and a critical point corresponding to a hard disk steady-state operating temperature that has the smallest difference from the hard disk safe operating temperature upper limit is taken as the second segment threshold.

Images