Method and device for controlling operation of RAID (redundant array of independent disks) card
By using the target prediction network in the RAID card to predict the operation indicators of the RAID card, the timely control problem caused by the periodic detection of the RAID card operation indicators is solved, and timely prediction and control of the operation of the RAID card is achieved, and the system stability is improved.
Patent Information
- Application Number
- CN202411046203.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-31
- Publication Date
- 2024-11-26
AI Technical Summary
In the prior art, the RAID card operation indicator detection is periodic, resulting in the inability to trigger control measures in time, resulting in the problem of processor restart or server downtime.
By obtaining current and historical RAID card operation indicators, using the target prediction network to predict, calculate smooth parameters and trend parameters, predict the target operation indicators of RAID card, and control the operation of RAID card based on the predictive indicators, including real-time monitoring of temperature and load and Adjustment.
It realizes timely prediction and control of RAID card operation indicators, avoids the risks of processor restart and server downtime, and improves the stability and reliability of the system.
Smart Images

Figure CN119025031A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of computers, and more particularly, to a method and apparatus for controlling the operation of a disk array (RAID) card. Background Art
[0002] When the Redundant Array of lndependent Disks (RAID) card is in a high-speed operation state, the processor load, temperature and other RAID card operating indicators rise sharply. If the temperature control system is not designed properly, it will cause the RAID card operating indicators to rise rapidly above the alarm threshold, causing the processor to restart or even the server to crash. Generally, the acquisition unit on the RAID card periodically detects the RAID card operating indicators to determine whether the operating indicators exceed the set threshold. If the threshold is exceeded, measures are triggered to control the operation of the disk array RAID card; if the threshold is not exceeded, continue to wait for the next cycle. Therefore, there is a situation where the RAID card operating indicator is not detected to exceed the threshold during the detection cycle, and the RAID card operating indicator is detected to exceed the threshold after the detection cycle, resulting in the problem of not triggering measures to control the operation of the disk array RAID card in time. Summary of the Invention
[0003] The embodiments of the present application provide a method and apparatus for controlling the operation of a disk array RAID card, so as to at least solve the problem in the related art that the operation of the RAID card cannot be controlled in a timely manner.
[0004] According to one embodiment of the present application, a method for controlling the operation of a disk array (RAID) card is provided, which is applied to an operation control device, wherein the operation control device is deployed in a RAID card. The method includes: when it is determined that the RAID card is in an operating state at a current time, obtaining the current operating index of the RAID card and N historical operating indexes at N historical times, wherein the N historical times are times before the current time, and the N is a natural number greater than or equal to 1; inputting the current operating index and the N historical operating indexes into a target prediction network to obtain a target operating index, wherein the target prediction network is obtained by training an initial prediction network using multiple groups of training operating index sequences, each group of the training operating index sequences includes multiple training operating indexes sorted by time, the initial prediction network includes a calculation subnetwork and a prediction subnetwork, for each group of the training operating index sequences, the calculation subnetwork is used to calculate a smoothing parameter between the multiple training operating indicators, and the prediction subnetwork is used to predict the operating index based on the smoothing parameter; and controlling the operation of the RAID card based on the target operating index.
[0005] In an exemplary embodiment, the operation control device includes an indicator collection service. When it is determined that the RAID card is in operation at the current time, the current operation indicator of the RAID card and N historical operation indicators at N historical times are obtained, including: setting an indicator collection period for the indicator collection service; collecting the operation indicator of the RAID card according to the indicator collection period through the indicator collection service to obtain the current operation indicator and N historical operation indicators. When a processor and memory are provided in the RAID card, the current operation indicator and the historical operation indicator both include the temperature information of the RAID card and the load information of the processor, and the current operation indicator and the historical operation indicator are both stored in the memory. The processor is used to process read and write operations in the RAID card.
[0006] In an exemplary embodiment, the initial prediction network is trained using multiple sets of training operation indicator sequences, including: iteratively training the initial prediction network using the above-mentioned multiple sets of training operation indicator sequences until the number of times the initial prediction network is trained reaches a target number, or when the loss value output by the target loss function of the above-mentioned initial prediction network meets a preset training end condition, ending the training to obtain the above-mentioned target prediction network; wherein the loss value output by the above-mentioned target loss function is determined based on the loss value output by the first loss function of the above-mentioned computing subnetwork and the loss value output by the second loss function of the above-mentioned prediction subnetwork, and the loss value output by the above-mentioned first loss function is related to the smoothing parameter output by the above-mentioned computing subnetwork.
[0007] In an exemplary embodiment, the above-mentioned multiple sets of training operation indicator sequences are used to iteratively train the above-mentioned initial prediction network, including: performing the i-th training on the above-mentioned initial prediction network through the following steps, wherein the above-mentioned i is a positive integer: obtaining the i-th set of training operation indicator sequences used for the i-th training of the above-mentioned initial prediction network from the above-mentioned multiple sets of training operation indicator sequences; inputting the multiple training operation indicators in the above-mentioned training operation indicator sequences into the computing subnetwork trained for the i-1th time, and obtaining smoothing parameters between the multiple training operation indicators output by the above-mentioned computing subnetwork, wherein the above-mentioned smoothing parameters include horizontal smoothing parameters, trend smoothing parameters and seasonal smoothing parameters; inputting the smoothing parameters between the multiple training operation indicators into the above-mentioned In the prediction subnetwork, the predicted operation index of the i-th training is obtained; the first loss value output by the above-mentioned first loss function of the above-mentioned i-th training is determined according to the smoothing parameter between the multiple above-mentioned training operation indicators output by the above-mentioned calculation subnetwork, and the second loss value output by the above-mentioned second loss function is determined according to the predicted operation index output by the above-mentioned prediction subnetwork of the above-mentioned i-th training and the actual operation index; the target loss value output by the above-mentioned target loss function is determined based on the above-mentioned first loss value and the above-mentioned second loss value; when i is less than the above-mentioned target number of times, or when the above-mentioned target loss value does not meet the above-mentioned training end condition, the value of the smoothing parameter in the above-mentioned initial prediction network of the above-mentioned i-1-th training is adjusted to obtain the above-mentioned initial prediction network of the above-mentioned i-th training.
[0008] In an exemplary embodiment, the formula included in the calculation subnetwork for calculating the smoothing parameter is:
[0009] s i =α(x i -p i-k )+(1-α)(s i-1 +t i-1 );
[0010] t i =β(s i -s i-1 )+(1-β)t i-1 ;
[0011] p i =γ(x i -s i )+(1-γ)p i-k ; Among them, the above s i It is used to represent the smoothing value between the multiple training run indicators in the above i-th training. i It is used to represent the trend value between the multiple training operation indicators in the above i-th training. The above p iIt is used to represent the seasonal value between the multiple training operation indicators in the above i-th training, the above α is used to represent the horizontal smoothing parameter between the multiple training operation indicators in the i-1-th training, the above β is used to represent the trend smoothing parameter between the multiple training operation indicators in the i-1-th training, the above γ is used to represent the period smoothing parameter between the multiple training operation indicators in the i-1-th training, the above k is used to represent the length of the seasonal cycle, the above h is used to represent the time length between the forecast period and the above i, and the above x i is the actual running indicator during the i-th training.
[0012] In an exemplary embodiment, the formula included in the prediction subnetwork for predicting the above-mentioned prediction operation index is: i+h =s i +ht i +p i-k+h ; Among them, the above x i+h is the above-mentioned prediction operation index during the i-th training.
[0013] In an exemplary embodiment, the RAID card is connected to a baseboard management controller, and the operation control device further includes an indicator control service, which controls the operation of the RAID card based on the target operation indicator, including: when the temperature operation indicator in the target operation indicator exceeds a first preset threshold, sending a first control instruction to the baseboard management controller through the indicator control service to control the operation of the temperature control device through the baseboard management controller, wherein the temperature control device is used to control the operating temperature of the RAID card; when the load operation indicator in the target operation indicator exceeds a second preset threshold, sending a second control instruction to the processor set in the RAID card through the indicator control service to control the load of the processor control operation.
[0014] According to another embodiment of the present application, a device for controlling the operation of a disk array (RAID) card is provided, which is applied to an operation control device, wherein the operation control device is deployed in the RAID card, and the device includes: a first acquisition module, for obtaining the current operation index of the RAID card and N historical operation indexes at N historical times, when it is determined that the RAID card is in an operation state at the current time, wherein the N historical times are times before the current time, and N is a natural number greater than or equal to 1; a first input module, for inputting the current operation index and the N historical operation indexes into a target prediction network to obtain a target operation index, wherein the target prediction network is obtained by training an initial prediction network using multiple sets of training operation index sequences, each set of the training operation index sequences includes multiple training operation indexes sorted by time, the initial prediction network includes a calculation subnetwork and a prediction subnetwork, for each set of the training operation index sequences, the calculation subnetwork is used to calculate a smoothing parameter between the multiple training operation indexes, and the prediction subnetwork is used to predict the operation index based on the smoothing parameter; and a first control module, for controlling the operation of the RAID card based on the target operation index.
[0015] In an exemplary embodiment, the above-mentioned operation control device includes an indicator collection service, and the above-mentioned first acquisition module includes: a first setting submodule, used to set the indicator collection period of the above-mentioned indicator collection service; a first collection submodule, used to collect the operation indicators of the above-mentioned RAID card according to the above-mentioned indicator collection period through the above-mentioned indicator collection service, and obtain the above-mentioned current operation indicators and N above-mentioned historical operation indicators, wherein, when the above-mentioned RAID card is provided with a processor and memory, the above-mentioned current operation indicators and the above-mentioned historical operation indicators both include the temperature information of the above-mentioned RAID card and the load information of the above-mentioned processor, and the above-mentioned current operation indicators and the above-mentioned historical operation indicators are both stored in the above-mentioned memory, and the above-mentioned processor is used to process the read and write operations in the above-mentioned RAID card.
[0016] In an exemplary embodiment, the initial prediction network is trained using multiple sets of training operation indicator sequences, including: iteratively training the initial prediction network using the above-mentioned multiple sets of training operation indicator sequences until the number of times the initial prediction network is trained reaches a target number, or when the loss value output by the target loss function of the above-mentioned initial prediction network meets a preset training end condition, ending the training to obtain the above-mentioned target prediction network; wherein the loss value output by the above-mentioned target loss function is determined based on the loss value output by the first loss function of the above-mentioned computing subnetwork and the loss value output by the second loss function of the above-mentioned prediction subnetwork, and the loss value output by the above-mentioned first loss function is related to the smoothing parameter output by the above-mentioned computing subnetwork.
[0017] In an exemplary embodiment, the above-mentioned multiple sets of training operation indicator sequences are used to iteratively train the above-mentioned initial prediction network, including: performing the i-th training on the above-mentioned initial prediction network through the following steps, wherein the above-mentioned i is a positive integer: obtaining the i-th set of training operation indicator sequences used for the i-th training of the above-mentioned initial prediction network from the above-mentioned multiple sets of training operation indicator sequences; inputting the multiple training operation indicators in the above-mentioned training operation indicator sequences into the computing subnetwork of the i-1-th training, and obtaining smoothing parameters between the multiple training operation indicators output by the above-mentioned computing subnetwork, wherein the above-mentioned smoothing parameters include horizontal smoothing parameters, trend smoothing parameters and period smoothing parameters; inputting the smoothing parameters between the multiple training operation indicators into the above-mentioned In the prediction subnetwork, the predicted operation index of the i-th training is obtained; the first loss value output by the above-mentioned first loss function of the above-mentioned i-th training is determined according to the smoothing parameter between the multiple above-mentioned training operation indicators output by the above-mentioned calculation subnetwork, and the second loss value output by the above-mentioned second loss function is determined according to the predicted operation index output by the above-mentioned prediction subnetwork of the above-mentioned i-th training and the actual operation index; the target loss value output by the above-mentioned target loss function is determined based on the above-mentioned first loss value and the above-mentioned second loss value; when i is less than the above-mentioned target number of times, or when the above-mentioned target loss value does not meet the above-mentioned training end condition, the value of the smoothing parameter in the above-mentioned initial prediction network of the above-mentioned i-1-th training is adjusted to obtain the above-mentioned initial prediction network of the above-mentioned i-th training.
[0018] In an exemplary embodiment, the formula included in the calculation subnetwork for calculating the smoothing parameter is:
[0019] s i =α(x i -p i-k )+(1-α)(s i-1 +t i-1 );
[0020] t i =β(s i -s i-1 )+(1-β)t i-1 ;
[0021] p i =γ(x i -s i )+(1-γ)p i-k ; Among them, the above s i It is used to represent the smoothing value between the multiple training run indicators in the above i-th training. i It is used to represent the trend value between the multiple training operation indicators in the above i-th training. The above p iIt is used to represent the seasonal value between the multiple training operation indicators in the above i-th training, the above α is used to represent the horizontal smoothing parameter between the multiple training operation indicators in the i-1-th training, the above β is used to represent the trend smoothing parameter between the multiple training operation indicators in the i-1-th training, the above γ is used to represent the period smoothing parameter between the multiple training operation indicators in the i-1-th training, the above k is used to represent the length of the seasonal cycle, the above h is used to represent the time length between the forecast period and the above i, and the above x i is the actual running indicator during the i-th training.
[0022] In an exemplary embodiment, the formula included in the prediction subnetwork for predicting the above-mentioned prediction operation index is: i+h =s i +ht i +p i-k+h ; Among them, the above x i+h is the above-mentioned prediction operation index during the i-th training.
[0023] In an exemplary embodiment, the RAID card is connected to a baseboard management controller, the operation control device further includes an indicator control service, and the first control module includes: a first sending submodule for sending a first control instruction to the baseboard management controller through the indicator control service when the temperature operation indicator in the target operation indicator exceeds a first preset threshold value, so as to control the operation of the temperature control device through the baseboard management controller, wherein the temperature control device is used to control the operating temperature of the RAID card; a second sending submodule for sending a second control instruction to the processor set in the RAID card through the indicator control service when the load operation indicator in the target operation indicator exceeds a second preset threshold value, so as to control the load controlled by the processor.
[0024] According to another embodiment of the present application, a computer-readable storage medium is provided, in which a computer program is stored. The computer program is configured to execute the steps of any one of the above method embodiments when running.
[0025] According to another embodiment of the present application, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.
[0026] According to another embodiment of the present application, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the steps in any one of the above method embodiments are implemented.
[0027] This application uses a target prediction network to predict the target operating indicators of a RAID card at a target time based on the current operating indicators of the RAID card and N historical operating indicators at N historical times. The target operating indicators are then used to control the operation of the RAID card. Because this embodiment predicts the target operating indicators of the RAID card at the target time, it can address the problem of related art in being unable to timely control the operation of the RAID card. This achieves the effect of predicting the RAID card operating indicators and timely controlling the operation of the RAID card. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] Figure 1 This is a hardware structure block diagram of a server device for a method of controlling the operation of a disk array RAID card according to an embodiment of the present application;
[0029] Figure 2 is a flow chart of a method for controlling the operation of a disk array RAID card according to an embodiment of the present application;
[0030] Figure 3 This is a structural diagram of a method for controlling the operation of a disk array RAID card according to a specific embodiment of the present application;
[0031] Figure 4 is a schematic diagram of a method for controlling the operation of a disk array RAID card according to a specific embodiment of the present application;
[0032] Figure 5 This is a flow chart of a method for controlling the operation of a disk array RAID card according to a specific embodiment of the present application;
[0033] Figure 6 This is a structural block diagram of a device for controlling the operation of a disk array RAID card according to an embodiment of the present application. DETAILED DESCRIPTION
[0034] The embodiments of the present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.
[0035] It should be noted that the terms "first", "second", etc. in the description and claims of this application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.
[0036] The method embodiments provided in the embodiments of the present application can be executed in a server device or a similar computing device. Taking running on a server device as an example, Figure 1 This is a hardware structure diagram of a server device for a method of controlling the operation of a disk array RAID card according to an embodiment of the present application. Figure 1 As shown, the server device may include one or more ( Figure 1Only one is shown) a processor 102 (the processor 102 may include but is not limited to a microprocessor MCU or a programmable logic device FPGA and other processing devices) and a memory 104 for storing data, wherein the above-mentioned server device may also include a transmission device 106 for communication functions and an input and output device 108. It will be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above server device. Figure 1 More or fewer components than shown, or with Figure 1 Different configurations shown.
[0037] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as a computer program corresponding to a method for controlling the operation of a disk array RAID card in an embodiment of the present application. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, implementing the above-mentioned method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include a memory remotely located relative to the processor 102, and these remote memories may be connected to a server device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
[0038] The transmission device 106 is used to receive or transmit data via a network. A specific example of the aforementioned network may include a wireless network provided by a communication provider of a server device. In one embodiment, the transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to enable communication with the Internet. In another embodiment, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.
[0039] In this embodiment, a method for controlling the operation of a disk array RAID card is provided, which is applied to an operation control device. The operation control device is deployed in the RAID card. Figure 2 FIG. 1 is a flow chart of a method for controlling the operation of a disk array RAID card according to an embodiment of the present application. Figure 2 As shown, the process includes the following steps:
[0040] Step S202: If it is determined that the RAID card is in an operating state at the current time, obtain the current operating index of the RAID card and N historical operating indexes at N historical times, where the N historical times are times before the current time, and N is a natural number greater than or equal to 1.
[0041] Step S204: Input the current operating indicator and the N historical operating indicators into a target prediction network to obtain a target operating indicator, wherein the target prediction network is obtained by training an initial prediction network using multiple sets of training operating indicator sequences, each set of the training operating indicator sequences including multiple training operating indicators sorted by time, and the initial prediction network including a calculation subnetwork and a prediction subnetwork. For each set of the training operating indicator sequences, the calculation subnetwork is used to calculate a smoothing parameter between the multiple training operating indicators, and the prediction subnetwork is used to predict the operating indicator based on the smoothing parameter.
[0042] Step S206: Control the operation of the RAID card based on the target operation index.
[0043] Alternatively, you can determine whether the RAID card is in a running state at the current time through hardware status indicators, system logs, or management software.
[0044] Optionally, the operating indicators include but are not limited to temperature information of the RAID card, load information of the processor, and power usage information. The collected current and historical operating indicators can be cleaned and formatted to meet the input requirements of the target prediction network.
[0045] Optionally, the computing subnetwork may utilize a convolutional neural network for computing, and the prediction subnetwork may utilize a recurrent neural network or a long short-term memory network to process time and predict operating indicators based on smoothing parameters.
[0046] Optionally, the smoothing parameters include but are not limited to horizontal smoothing parameters, trend smoothing parameters, and seasonal smoothing parameters.
[0047] Through the above steps, based on the current operating indicators of the RAID card and N historical operating indicators at N historical times, the target prediction network is used to predict the target operating indicators of the RAID card at a target time, and the operation of the RAID card is controlled based on the target operating indicators. Because this embodiment predicts the target operating indicators of the RAID card at the target time, it can solve the problem of the related art of being unable to timely control the operation of the RAID card, and achieves the effect of predicting the RAID card operating indicators and timely controlling the operation of the RAID card.
[0048] In an exemplary embodiment, the operation control device includes an indicator collection service. When it is determined that the RAID card is in operation at the current time, the current operation indicator of the RAID card and N historical operation indicators at N historical times are obtained, including: setting an indicator collection period for the indicator collection service; collecting the operation indicator of the RAID card according to the indicator collection period through the indicator collection service to obtain the current operation indicator and N historical operation indicators. When a processor and memory are provided in the RAID card, the current operation indicator and the historical operation indicator both include the temperature information of the RAID card and the load information of the processor, and the current operation indicator and the historical operation indicator are both stored in the memory. The processor is used to process read and write operations in the RAID card.
[0049] Optionally, the indicator collection period can be in seconds, for example, 2 seconds, 3 seconds, or 5 seconds. Figure 3 As shown, Figure 3 It is a structural diagram of a method for controlling the operation of a disk array RAID card according to a specific embodiment of the present application. The indicator collection service in the operation control device collects operation indicators every 2 seconds: the temperature information of the RAID card and the load information of the processor. As shown in Table 1, at 10:00:00 on 2020-07-26, the temperature information of the RAID card is 35°C, and the load information of the processor is 25%; at 10:00:02 on 2020-07-26, the temperature information of the RAID card is 37°C, and the load information of the processor is 45%; at 10:00:04 on 2020-07-26, the temperature information of the RAID card is 36°C, and the load information of the processor is 30%; at 10:00:06 on 2020-07-26, the temperature information of the RAID card is 38°C, and the load information of the processor is 60%; at 10:00:08 on 2020-07-26, the temperature information of the RAID card is 39°C, and the load information of the processor is 50%. This embodiment achieves the purpose of monitoring the operating status of the RAID card by setting an indicator collection service in the operation control device to regularly collect operating indicators.
[0050] Table 1:
[0051] time RAID card temperature information (°C) Processor load information (%) 2020-07-26 10:00:00 35 25 2020-07-26 10:00:02 37 45 2020-07-26 10:00:04 36 30 2020-07-26 10:00:06 38 60 2020-07-26 10:00:08 39 50
[0052] In an exemplary embodiment, the initial prediction network is trained using multiple sets of training operation indicator sequences, including: iteratively training the initial prediction network using the above-mentioned multiple sets of training operation indicator sequences until the number of times the initial prediction network is trained reaches a target number, or when the loss value output by the target loss function of the above-mentioned initial prediction network meets a preset training end condition, ending the training to obtain the above-mentioned target prediction network; wherein the loss value output by the above-mentioned target loss function is determined based on the loss value output by the first loss function of the above-mentioned computing subnetwork and the loss value output by the second loss function of the above-mentioned prediction subnetwork, and the loss value output by the above-mentioned first loss function is related to the smoothing parameter output by the above-mentioned computing subnetwork.
[0053] Alternatively, multiple sets of training operating indicator sequences can be obtained by collecting temperature information from different RAID cards under different workloads, environments, and humidity conditions, or by collecting load information from different RAID cards under different temperature and humidity conditions. The loss values of the first and second loss functions can be measured using loss functions such as mean squared error, mean absolute error, and cross-entropy loss. For example, iterative training can be terminated when the mean squared error between the smoothing parameter output by the calculation subnetwork and the actual smoothing parameter is less than 0.05, and iterative training can be terminated when the mean squared error between the operating indicator output by the prediction subnetwork and the actual operating indicator is less than 0.06. The loss value output by the target loss function can be obtained using a weighted average method, combining the outputs of the first and second loss functions. This embodiment, through iterative training and loss function optimization, simultaneously integrates two independent loss functions to achieve a more robust and stable target prediction network, enabling more accurate prediction of RAID card operating indicators and reducing prediction errors. Furthermore, the training termination condition can be flexibly set according to actual needs, such as the number of training sessions or when the loss value meets specific conditions, making the training process more flexible and controllable.
[0054] In an exemplary embodiment, the above-mentioned multiple sets of training operation indicator sequences are used to iteratively train the above-mentioned initial prediction network, including: performing the i-th training on the above-mentioned initial prediction network through the following steps, wherein the above-mentioned i is a positive integer: obtaining the i-th set of training operation indicator sequences used for the i-th training of the above-mentioned initial prediction network from the above-mentioned multiple sets of training operation indicator sequences; inputting the multiple training operation indicators in the above-mentioned training operation indicator sequences into the computing subnetwork of the i-1-th training, and obtaining smoothing parameters between the multiple training operation indicators output by the above-mentioned computing subnetwork, wherein the above-mentioned smoothing parameters include horizontal smoothing parameters, trend smoothing parameters and period smoothing parameters; inputting the smoothing parameters between the multiple training operation indicators into the above-mentioned In the prediction subnetwork, the predicted operation index of the i-th training is obtained; the first loss value output by the above-mentioned first loss function of the above-mentioned i-th training is determined according to the smoothing parameter between the multiple above-mentioned training operation indicators output by the above-mentioned calculation subnetwork, and the second loss value output by the above-mentioned second loss function is determined according to the predicted operation index output by the above-mentioned prediction subnetwork of the above-mentioned i-th training and the actual operation index; the target loss value output by the above-mentioned target loss function is determined based on the above-mentioned first loss value and the above-mentioned second loss value; when i is less than the above-mentioned target number of times, or when the above-mentioned target loss value does not meet the above-mentioned training end condition, the value of the smoothing parameter in the above-mentioned initial prediction network of the above-mentioned i-1-th training is adjusted to obtain the above-mentioned initial prediction network of the above-mentioned i-th training.
[0055] Optionally, predicting the temperature information of the RAID card is taken as an example: the third set of training operation indicator sequences used for the third training is obtained from multiple sets of training operation indicator sequences: at 10:00:00 on 2020-06-26, the temperature information of the RAID card is 30°C; at 10:00:02 on 2020-06-26, the temperature information of the RAID card is 33°C; at 10:00:04 on 2020-06-26, the temperature information of the RAID card is 36°C; at 10:00:06 on 2020-06-26, the temperature information of the RAID card is 34°C; at 10:00:08 on 2020-06-26, the temperature information of the RAID card is 34°C. The third set of training performance indicators was input into the computational subnetwork from the second training run. The smoothing parameters for the temperature information output by the current computational subnetwork were obtained, including a horizontal smoothing parameter of 0.3, a trend smoothing parameter of 0.3, and a period smoothing parameter of 0.3. The prediction subnetwork used the smoothing parameters output by the computational subnetwork as input to generate the predicted performance indicators for the third training run: the RAID card temperature at 10:00:10 on June 26, 2020, was 33°C; and the RAID card temperature at 10:00:12 on June 26, 2020, was 30°C. The first loss between the smoothing parameters output by the computational subnetwork and the actual smoothing parameters was a mean squared error of 0.04. The second loss between the performance indicators output by the prediction subnetwork and the actual performance indicators was a mean squared error of 0.06. Combining the first and second loss values, a weighted average was used to determine the target loss function output of 0.05. The target loss value of 0.05 does not meet the preset training end condition of 0.04. Iterative training continues based on the initial prediction network obtained from the third training run until the preset training end condition is met. The output of the target loss function is determined using a weighted average, combining the first and second loss values. This balances the performance of the computational and prediction subnetworks, ensuring the coordination of the entire prediction network. This embodiment achieves the purpose of accurately determining the smoothing parameter by iteratively training the initial prediction network using multiple sets of training run indicator sequences.
[0056] In an exemplary embodiment, the formula included in the calculation subnetwork for calculating the smoothing parameter is:
[0057] s i =α(x i -p i-k )+(1-α)(s i-1 +t i-1 );
[0058] t i =β(s i -s i-1 )+(1-β)t i-1 ;
[0059] p i =γ(x i -s i )+(1-γ)p i-k ; Among them, the above s i It is used to represent the smoothing value between the multiple training run indicators in the above i-th training. i It is used to represent the trend value between the multiple training operation indicators in the above i-th training. The above p i It is used to represent the seasonal value between the multiple training operation indicators in the above i-th training, the above α is used to represent the horizontal smoothing parameter between the multiple training operation indicators in the i-1-th training, the above β is used to represent the trend smoothing parameter between the multiple training operation indicators in the i-1-th training, the above γ is used to represent the period smoothing parameter between the multiple training operation indicators in the i-1-th training, the above k is used to represent the length of the seasonal cycle, the above h is used to represent the time length between the forecast period and the above i, and the above x i is the actual running indicator during the i-th training.
[0060] In an exemplary embodiment, the formula included in the prediction subnetwork for predicting the above-mentioned prediction operation index is: i+h =s i +ht i +p i-k+h ; Among them, the above x i+h is the above-mentioned prediction operation index during the i-th training.
[0061] Alternatively, take the training run indicators x1 = 20, x2 = 22, x3 = 24, x4 = 23, x5 = 27, smoothing values s0 = 20, t0 = 2, p0 = 20, and the initial α = 0.3, β = 0.3, γ = 0.3, k = 4, and h = 1 for the 0th time as an example for illustration: s1 = 0.3 (20) + 0.7 (20 + 2) = 6 + 14.8 = 20.8, t1 = 0.3 (20.8 - 20) + 0.7 2 = 0.24 + 1.4 = 1. 64, p1 = 20; s2 = 0.3.(22-20) + 0.7 (20.8 + 1.64) = 0.6 + 14.476 = 15.076, t2 = 0.3 (15.076-20.8) + 0.7 1.64 = -1.0248 + 1.148 = 0.1232, p2 = 0.3 (22-15.076) + 0.7 20 = 1.7364 + 14 = 15.7364, repeat the calculation of s3, t3, p3; t i, t4, p4. Assuming that s3 = 23.24, t4 = 0.48, and p4 = 23 have been calculated, then x5 = 23.24 + 10.48 + 23 = 46.72, which clearly does not conform to the actual situation. This indicates that the smoothing parameter is inaccurate. The initial prediction network can be iteratively trained using more training run indicator sequences. This embodiment iteratively trains the initial prediction network using multiple sets of training run indicator sequences, while minimizing the difference between the predicted values and the actual values, continuously optimizing and reducing the prediction error, thereby achieving the purpose of accurately determining the smoothing parameter.
[0062] In an exemplary embodiment, the RAID card is connected to a baseboard management controller, and the operation control device further includes an indicator control service, which controls the operation of the RAID card based on the target operation indicator, including: when the temperature operation indicator in the target operation indicator exceeds a first preset threshold, sending a first control instruction to the baseboard management controller through the indicator control service to control the operation of the temperature control device through the baseboard management controller, wherein the temperature control device is used to control the operating temperature of the RAID card; when the load operation indicator in the target operation indicator exceeds a second preset threshold, sending a second control instruction to the processor set in the RAID card through the indicator control service to control the load of the processor control operation.
[0063] Optionally, the temperature control device includes but is not limited to a cooling fan. Assuming that the smoothing parameters obtained based on training are α=0.2, β=0.4, and γ=0.3, based on the smoothing parameters and s i =α(x i -p i-k )+(1-α)(s i-1 +t i-1 );t i =β(s i -s i-1 )+(1-β)t i-1 ;p i =γ(x i -s i )+(1-γ)p i-k ;x i+h =s i +ht i +p i-k+hThe target operating indicators for the future time are obtained: the temperature information at 10:01:00 on 2020-06-26 is 33°C, the load information is 35%, the temperature information at 10:01:02 on 2020-06-26 is 37°C, the load information is 35%, the temperature information at 10:01:04 on 2020-06-26 is 33°C, the load information is 30%, and the temperature at 10:01:00 on 2020-06-26 exceeds the first preset threshold of 32°C, and the load information exceeds the second threshold of 34%. Figure 3 As shown, Figure 3 This is a block diagram of a method for controlling the operation of a disk array (RAID) card according to a specific embodiment of the present application. The indicator control service in the operation control device sends a first control instruction to the baseboard management controller, instructing the cooling fan to cool the RAID card before 10:01:02 on June 26, 2020. The indicator control service then sends a second control instruction to the business service in the processor provided in the RAID card to control the processor's control operation load before 10:01:02 on June 26, 2020. This embodiment achieves the purpose of timely controlling the operation of the RAID card by predicting the RAID card's operating indicators.
[0064] The present application is described below with reference to specific embodiments:
[0065] Figure 4 FIG. 1 is a schematic diagram of a method for controlling the operation of a disk array RAID card according to a specific embodiment of the present application, such as Figure 4 As shown, this specific embodiment is explained with temperature detection. T3 is the current time, and T0, T1, and T2 are historical times. At T3, the RAID card indicator collection service reads the temperature information. At this time, the temperature of T3 is lower than the first threshold. The RAID card instructs the baseboard management controller to take cooling measures. The cooling measures are not triggered until T4 arrives. However, between T3 and T4, the temperature of T3' has reached the first threshold, so the RAID card does not take timely cooling measures during the time of T3'-T4. Therefore, s i =α(x i -p i-k )+(1-α)(s i-1 +t i-1 );t i =β(s i -s i-1 )+(1-β)t i-1 ;p i =γ(x i -s i )+(1-γ)p i-k ;x i+h =s i +ht i +p i-k+hThe temperature value of T4 is predicted at T3. If the predicted value exceeds a first threshold, a cooling measure is immediately triggered.
[0066] In this specific embodiment, Figure 5 FIG. 1 is a flow chart of a method for controlling the operation of a disk array RAID card according to a specific embodiment of the present application. Figure 5 As shown, the specific steps include:
[0067] S502, the RAI D card starts the indicator collection service to periodically collect temperature information and processor load information;
[0068] S504, after reaching the detection period;
[0069] S506, using s i =α(x i -p i-k )+(1-α)(s i-1 +t i-1 );t i =β(s i -s f-1 )+(1-β)t i-1 ;p i =γ(x i -s i )+(1-γ)p i-k ;x i+h =s i +ht i +p i-k+h Predict the temperature information and processor load information at the next detection time. If they exceed the set threshold, execute S508; otherwise, execute S504.
[0070] S508: When the temperature exceeds the high temperature threshold, the RAID card notifies the baseboard management controller to increase the fan speed. When the processor load information exceeds the alarm threshold, the RAID card actively reduces the proportion of unnecessary background tasks.
[0071] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus the necessary general hardware platform, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of the present application.
[0072] This embodiment also provides a device for controlling the operation of a disk array (RAID) card, which is applied to an operation control device. The operation control device is deployed in the RAID card. This device is used to implement the above-mentioned embodiments and preferred embodiments, and the details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation using hardware, or a combination of software and hardware, is also possible and contemplated.
[0073] Figure 6 This is a structural block diagram of a device for controlling the operation of a disk array RAID card according to an embodiment of the present application, such as Figure 6 As shown, the device includes:
[0074] A first acquisition module 602 is configured to, upon determining that the RAID card is in an operating state at a current time, acquire a current operating indicator of the RAID card and N historical operating indicators at N historical times, where the N historical times are times before the current time, and N is a natural number greater than or equal to 1;
[0075] A first input module 604 is configured to input the current operating indicator and the N historical operating indicators into a target prediction network to obtain a target operating indicator, wherein the target prediction network is obtained by training an initial prediction network using multiple sets of training operating indicator sequences, each set of the training operating indicator sequences including multiple training operating indicators sorted by time, the initial prediction network including a calculation subnetwork and a prediction subnetwork, and for each set of the training operating indicator sequences, the calculation subnetwork is configured to calculate a smoothing parameter between the multiple training operating indicators, and the prediction subnetwork is configured to predict the operating indicator based on the smoothing parameter;
[0076] The first control module 606 is configured to control the operation of the RAID card based on the target operation index.
[0077] In an exemplary embodiment, the above-mentioned operation control device includes an indicator collection service, and the above-mentioned first acquisition module includes: a first setting submodule, used to set the indicator collection period of the above-mentioned indicator collection service; a first collection submodule, used to collect the operation indicators of the above-mentioned RA ID card according to the above-mentioned indicator collection period through the above-mentioned indicator collection service, and obtain the above-mentioned current operation indicators and N above-mentioned historical operation indicators, wherein, when the above-mentioned RAID card is provided with a processor and memory, the above-mentioned current operation indicators and the above-mentioned historical operation indicators both include the temperature information of the above-mentioned RAID card and the load information of the above-mentioned processor, and the above-mentioned current operation indicators and the above-mentioned historical operation indicators are both stored in the above-mentioned memory, and the above-mentioned processor is used to process the read and write operations in the above-mentioned RAID card.
[0078] In an exemplary embodiment, the initial prediction network is trained using multiple sets of training operation indicator sequences, including: iteratively training the initial prediction network using the above-mentioned multiple sets of training operation indicator sequences until the number of times the initial prediction network is trained reaches a target number, or when the loss value output by the target loss function of the above-mentioned initial prediction network meets a preset training end condition, ending the training to obtain the above-mentioned target prediction network; wherein the loss value output by the above-mentioned target loss function is determined based on the loss value output by the first loss function of the above-mentioned computing subnetwork and the loss value output by the second loss function of the above-mentioned prediction subnetwork, and the loss value output by the above-mentioned first loss function is related to the smoothing parameter output by the above-mentioned computing subnetwork.
[0079] In an exemplary embodiment, the above-mentioned multiple sets of training operation indicator sequences are used to iteratively train the above-mentioned initial prediction network, including: performing the i-th training on the above-mentioned initial prediction network through the following steps, wherein the above-mentioned i is a positive integer: obtaining the i-th set of training operation indicator sequences used for the i-th training of the above-mentioned initial prediction network from the above-mentioned multiple sets of training operation indicator sequences; inputting the multiple training operation indicators in the above-mentioned training operation indicator sequences into the computing subnetwork of the i-1-th training, and obtaining smoothing parameters between the multiple training operation indicators output by the above-mentioned computing subnetwork, wherein the above-mentioned smoothing parameters include horizontal smoothing parameters, trend smoothing parameters and period smoothing parameters; inputting the smoothing parameters between the multiple training operation indicators into the above-mentioned In the prediction subnetwork, the predicted operation index of the i-th training is obtained; the first loss value output by the above-mentioned first loss function of the above-mentioned i-th training is determined according to the smoothing parameter between the multiple above-mentioned training operation indicators output by the above-mentioned calculation subnetwork, and the second loss value output by the above-mentioned second loss function is determined according to the predicted operation index output by the above-mentioned prediction subnetwork of the above-mentioned i-th training and the actual operation index; the target loss value output by the above-mentioned target loss function is determined based on the above-mentioned first loss value and the above-mentioned second loss value; when i is less than the above-mentioned target number of times, or when the above-mentioned target loss value does not meet the above-mentioned training end condition, the value of the smoothing parameter in the above-mentioned initial prediction network of the above-mentioned i-1-th training is adjusted to obtain the above-mentioned initial prediction network of the above-mentioned i-th training.
[0080] In an exemplary embodiment, the formula included in the calculation subnetwork for calculating the smoothing parameter is:
[0081] s i =α(x i -p i-k )+(1-α)(s i-1 +t i-1 );
[0082] t i =β(s i -s i-1 )+(1-β)t i-1 ;
[0083] p i =γ(x i -s i )+(1-γ)p i-k ; Among them, the above s i It is used to represent the smoothing value between the multiple training run indicators in the above i-th training. i It is used to represent the trend value between the multiple training operation indicators in the above i-th training. The above p i It is used to represent the seasonal value between the multiple training operation indicators in the above i-th training, the above α is used to represent the horizontal smoothing parameter between the multiple training operation indicators in the i-1-th training, the above β is used to represent the trend smoothing parameter between the multiple training operation indicators in the i-1-th training, the above γ is used to represent the period smoothing parameter between the multiple training operation indicators in the i-1-th training, the above k is used to represent the length of the seasonal cycle, the above h is used to represent the time length between the forecast period and the above i, and the above x i is the actual running indicator during the i-th training.
[0084] In an exemplary embodiment, the formula included in the prediction subnetwork for predicting the above-mentioned prediction operation index is: i+h =s i +ht i +p i-k+h ; Among them, the above x i+h is the above-mentioned prediction operation index during the i-th training.
[0085] In an exemplary embodiment, the above-mentioned RAID card is connected to the baseboard management controller, and the above-mentioned operation control device also includes an indicator control service. The above-mentioned first control module includes: a first sending submodule that sends a first control instruction to the above-mentioned baseboard management controller through the above-mentioned indicator control service when the temperature operation indicator in the above-mentioned target operation indicator exceeds a first preset threshold value, so as to control the operation of the temperature control device through the above-mentioned baseboard management controller, wherein the above-mentioned temperature control device is used to control the operating temperature of the above-mentioned RAID card; a second sending submodule that is used to send a second control instruction to the processor set in the above-mentioned RAID card through the above-mentioned indicator control service when the load operation indicator in the above-mentioned target operation indicator exceeds a second preset threshold value, so as to control the load of the above-mentioned processor control operation.
[0086] It should be noted that the above modules can be implemented through software or hardware. For the latter, it can be implemented in the following ways, but not limited to: the above modules are all located in the same processor; or the above modules are located in different processors in any combination.
[0087] An embodiment of the present application further provides a computer-readable storage medium, in which a computer program is stored. The computer program is configured to execute the steps of any of the above method embodiments when run.
[0088] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk, or an optical disk.
[0089] An embodiment of the present application further provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.
[0090] In an exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.
[0091] An embodiment of the present application further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the steps of any of the above method embodiments are implemented.
[0092] An embodiment of the present application further provides another computer program product, comprising a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of any of the above method embodiments are implemented.
[0093] An embodiment of the present application further provides a computer program, which includes computer instructions stored in a computer-readable storage medium; a processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the steps of any of the above method embodiments.
[0094] For specific examples in this embodiment, reference may be made to the examples described in the above embodiments and exemplary implementation modes, and this embodiment will not be described in detail here.
[0095] Obviously, those skilled in the art should understand that the modules or steps of the present application described above can be implemented using a general-purpose computing device, they can be concentrated on a single computing device, or distributed across a network composed of multiple computing devices, they can be implemented using program code executable by the computing device, and thus, they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be performed in a different order than herein, or they can be fabricated into separate integrated circuit modules, or multiple modules or steps can be fabricated into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.
[0096] The above description is merely a preferred embodiment of the present application and is not intended to limit the present application. Various modifications and variations are possible for those skilled in the art. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present application shall be included within the scope of protection of the present application.
Claims
1. A method for controlling the operation of a disk array RAID card, characterized in that: Applied to an operation control device, the operation control device is deployed in a RAID card, and the method includes: When it is determined that the RAID card is in operation at the current time, obtaining the current operation index of the RAID card and N historical operation indexes at N historical times, wherein the N historical times are times before the current time, and N is a natural number greater than or equal to 1; Input the current operating index and N historical operating indexes into a target prediction network to obtain a target operating index, wherein the target prediction network is obtained by training an initial prediction network using multiple sets of training operating index sequences, each set of the training operating index sequences includes multiple training operating indexes sorted by time, the initial prediction network includes a calculation subnetwork and a prediction subnetwork, for each set of the training operating index sequences, the calculation subnetwork is used to calculate a smoothing parameter between the multiple training operating indexes, and the prediction subnetwork is used to predict the operating index based on the smoothing parameter; The operation of the RAID card is controlled based on the target operation index.
2. The method according to claim 1, characterized in that The operation control device includes an indicator collection service, which obtains the current operation indicator of the RAID card and N historical operation indicators at N historical times when it is determined that the RAID card is in operation at the current time, including: Setting the indicator collection period of the indicator collection service; The operating indicators of the RAID card are collected by the indicator collection service according to the indicator collection period to obtain the current operating indicators and N historical operating indicators, wherein, in the case where a processor and a memory are provided in the RAID card, the current operating indicators and the historical operating indicators both include temperature information of the RAID card and load information of the processor, and the current operating indicators and the historical operating indicators are both stored in the memory, and the processor is used to process read and write operations in the RAID card.
3. The method according to claim 1, characterized in that The initial prediction network is trained using multiple sets of training run indicator sequences, including: Iteratively training the initial prediction network using the multiple sets of training operation indicator sequences until the number of times the initial prediction network is trained reaches a target number of times, or when the loss value output by the target loss function of the initial prediction network meets a preset training end condition, the training is terminated to obtain the target prediction network; Among them, the loss value output by the target loss function is determined according to the loss value output by the first loss function of the computing subnetwork and the loss value output by the second loss function of the prediction subnetwork, and the loss value output by the first loss function is related to the smoothing parameter output by the computing subnetwork.
4. The method according to claim 3, characterized in that Iteratively training the initial prediction network using the multiple sets of training operation indicator sequences includes: The initial prediction network is trained for the i-th time by the following steps, wherein i is a positive integer: Obtaining, from the plurality of sets of training operation indicator sequences, an i-th set of training operation indicator sequences used for performing an i-th training on the initial prediction network; Inputting a plurality of the training operation indicators in the training operation indicator sequence into the computing subnetwork of the i-1th training, and obtaining smoothing parameters between the plurality of the training operation indicators output by the computing subnetwork, wherein the smoothing parameters include a horizontal smoothing parameter, a trend smoothing parameter, and a period smoothing parameter; Inputting the smoothing parameters between the plurality of training operation indicators into the prediction subnetwork to obtain the prediction operation indicator of the i-th training; Determine a first loss value output by the first loss function of the i-th training according to a smoothing parameter between the plurality of training operation indicators output by the calculation subnetwork, and determine a second loss value output by the second loss function according to a predicted operation indicator and a real operation indicator output by the prediction subnetwork of the i-th training; Determine a target loss value output by the target loss function based on the first loss value and the second loss value; When i is less than the target number of times, or when the target loss value does not meet the training end condition, the value of the smoothing parameter in the initial prediction network of the i-1th training is adjusted to obtain the initial prediction network of the i-th training.
5. The method according to claim 4, characterized in that The formula included in the calculation subnetwork for calculating the smoothing parameter is: s i =α(x i -p i-k )+(1-α)(s i-1 +t i-1 ); t i =β(s f -s i-1 )+(1-β)t i-1 ; p i =γ(x i -s i )+(1-γ)p i-k ; Among them, the s i It is used to represent the smoothing value between the multiple training operation indicators during the i-th training. i is used to represent the trend value between the plurality of training operation indicators during the i-th training, the p i is used to represent the seasonal value between the multiple training operation indicators in the i-th training, the α is used to represent the horizontal smoothing parameter between the multiple training operation indicators in the i-1-th training, the β is used to represent the trend smoothing parameter between the multiple training operation indicators in the i-1-th training, the γ is used to represent the period smoothing parameter between the multiple training operation indicators in the i-1-th training, the k is used to represent the length of the seasonal cycle, the h is used to represent the time length between the prediction period and the i, and the x i is the actual running indicator during the i-th training.
6. The method according to claim 5, characterized in that The formula included in the prediction subnetwork for predicting the prediction operation index is: x i+h =s i +ht i +p i-k+h ; Among them, the x i+h is the predicted operation index during the i-th training.
7. The method according to claim 1, characterized in that The RAID card is connected to a baseboard management controller, and the operation control device further includes an index control service, which controls the operation of the RAID card based on the target operation index, including: When the temperature operating index in the target operating index exceeds a first preset threshold, sending a first control instruction to the baseboard management controller through the index control service, so as to control the operation of a temperature control device through the baseboard management controller, wherein the temperature control device is used to control the operating temperature of the RAID card; When the load operation index in the target operation index exceeds a second preset threshold, a second control instruction is sent to a processor provided in the RAID card through the index control service to control the load of the processor control operation.
8. A device for controlling the operation of a disk array RAID card, characterized in that: Applied to an operation control device, the operation control device is deployed in a RAID card, and the device includes: A first acquisition module, configured to acquire, when determining that the RAID card is in operation at the current time, a current operation index of the RAID card and N historical operation indexes at N historical times, wherein the N historical times are times before the current time, and N is a natural number greater than or equal to 1; A first input module is used to input the current operating index and N historical operating indexes into a target prediction network to obtain a target operating index, wherein the target prediction network is obtained by training an initial prediction network using multiple sets of training operating index sequences, each set of the training operating index sequences includes multiple training operating indexes sorted by time, the initial prediction network includes a calculation subnetwork and a prediction subnetwork, for each set of the training operating index sequences, the calculation subnetwork is used to calculate a smoothing parameter between the multiple training operating indicators, and the prediction subnetwork is used to predict the operating index based on the smoothing parameter; The first control module is used to control the operation of the RAID card based on the target operation index.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program implements the steps of the method described in any one of claims 1 to 7 when executed by a processor.
10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method described in any one of claims 1 to 7 are implemented.
