Multi-modal data fusion data center optimization method and system

Abstract

The invention discloses a data center optimization method and system for multi-modal data fusion, and relates to the field of optimization control, and the method comprises a data collection module which is used for collecting multi-type data corresponding to a data center, and obtaining multi-modal data corresponding to the data center; the data preprocessing module is used for performing data processing on the multi-modal data corresponding to the data center to obtain multi-modal reference data corresponding to the data center; the data analysis module is used for performing data analysis according to the multi-modal reference data corresponding to the data center to obtain a multi-modal data fusion result corresponding to the data center; and the optimization control module is used for performing optimization control according to the multi-modal data fusion result corresponding to the data center. The method and the device have the effects of improving the flexibility of the data center in task allocation optimization and the task processing efficiency of the data center.

Description

Technical Field

[0001] The present application relates to the field of optimization control, and particularly to a data center optimization method and system based on multi-modal data fusion. Background Art

[0002] With the rapid development of cloud computing, big data, and artificial intelligence, the scale and complexity of modern data centers have been continuously climbing. As the core hub of computing, storage, and network resources, the operating efficiency and stability of the data center directly affect the quality of enterprise services and operating costs.

[0003] In the prior art, the optimization of data centers is usually based on data from a single dimension. This fragmented analysis mode ignores the relationships between various devices. At the same time, traditional optimization often relies on offline data analysis and manual intervention, making it difficult to respond in a timely manner to changes in real-time operating conditions, and the system is unable to quickly adjust the task allocation strategy. When optimizing the task allocation in the data center in the above way, it will lead to resource blind spots in the task allocation process, and local resources cannot be fully utilized, resulting in the overall working efficiency of the data center not meeting the work requirements. Summary of the Invention

[0004] The purpose of the present invention is to provide a data center optimization method and system based on multi-modal data fusion to solve the problems raised in the above background art.

[0005] To achieve the above purpose, the present invention provides the following technical solution: A data center optimization system based on multi-modal data fusion, including: A data acquisition module: used to collect various types of data corresponding to the data center to obtain multi-modal data corresponding to the data center; A data preprocessing module: used to perform data processing on the multi-modal data corresponding to the data center to obtain multi-modal reference data corresponding to the data center; A data analysis module: used to perform data analysis based on the multi-modal reference data corresponding to the data center to obtain a multi-modal data fusion result corresponding to the data center; An optimization control module: used to perform optimization control based on the multi-modal data fusion result corresponding to the data center.

[0006] In a preferred embodiment of this solution, the specific implementation method of the data acquisition module is as follows: Obtain the historical operation data corresponding to each type of device through a data monitoring platform, where the device types include servers, storage devices, and network devices; The historical operation data of the server includes the CPU utilization rate, memory usage rate, disk I / O, and network bandwidth usage rate corresponding to each server; The historical operation data of the storage device includes the disk utilization rate, read / write rate, IOPS, and latency corresponding to each storage device; The historical operation data of the network device includes the bandwidth utilization rate, packet loss rate, latency, throughput, and error rate corresponding to each network device; Obtain the historical task data of the data center through the task log. The historical task data includes each task executed by the data center, the task number of each task, the task basic data, the task execution time, the task termination time, the task resource requirements, the server number, storage device number, and network device number corresponding to each task; Among them, the task resource requirements include CPU requirements, network requirements, and storage requirements; Among them, the task basic data includes the computational complexity of the task and the data scale of the task; Record the historical operation data and historical task data corresponding to each type of device as the multi-modal data corresponding to the data center.

[0007] In a preferred solution of this scheme, the specific execution method of the data preprocessing module is as follows: Establish a data extraction relationship between the data preprocessing module and the database, and extract the standard task execution duration corresponding to each task execution model stored in the database. The task execution model includes the basic task data and task resource requirements, and obtain the standard task execution duration corresponding to each task through the basic task data and task resource requirements corresponding to each task; Obtain the actual task execution duration corresponding to each task through the task execution time and task termination time corresponding to each task. Compare the actual task execution duration corresponding to each task with the standard task execution duration. If the actual task execution duration corresponding to a certain task is less than or equal to the standard task execution duration, it means that the task is executed normally. If the actual task execution duration corresponding to a certain task is greater than the standard task execution duration, it means that the task is executed abnormally. Count the normally executed tasks and abnormally executed tasks; Obtain the numbers corresponding to each server, each storage device, and each network device in the data center. Perform matching and screening through the server number, storage device number, and network device number corresponding to each task to obtain the tasks corresponding to each server, each storage device, and each network device in the data center; Perform time screening through the task execution time and task termination time of the tasks corresponding to each server, each storage device, and each network device in the data center to obtain the normally executed tasks and abnormally executed tasks corresponding to each server, each storage device, and each network device in each preset time period. Count the total number of executed tasks, the number of normally executed tasks, and the number of abnormally executed tasks corresponding to each server, each storage device, and each network device in each preset time period; Record the total number of execution tasks, the number of normally executed tasks, and the number of abnormally executed tasks corresponding to each server, each storage device, and each network device in each preset time period as the multi-modal reference data corresponding to the data center.

[0008] In a preferred solution of this scheme, the specific execution method of the data analysis module is as follows: The specific execution method of the data analysis module is as follows: Establish a data extraction relationship between the data analysis module and the database, and extract various fault types caused by resource allocation or task allocation stored in the database; Obtain the historical fault data corresponding to each numbered device through the fault and repair data log. The historical fault data includes each fault corresponding to each numbered device, the fault type corresponding to each fault, and the fault time point. Match and screen the fault types corresponding to each fault of each numbered device with various fault types caused by resource allocation or task allocation. Denote the fault types that match the fault types caused by resource allocation or task allocation as reference fault types. Screen out each fault corresponding to the reference fault type of each numbered device, and denote it as each reference fault corresponding to each numbered device. Screen out the fault degree and fault time point corresponding to each reference fault; Screen and obtain each reference fault and fault time point corresponding to each server, each storage device, and each network device through the numbers corresponding to each server, each storage device, and each network device; Screen through the fault time points corresponding to each reference fault of each server, each storage device, and each network device to obtain each normal execution task, each abnormal execution task, the total number of execution tasks, the number of normal execution tasks, the number of abnormal execution tasks, and the historical operation data corresponding to each server, each storage device, and each network device at the fault time points corresponding to each reference fault; Through the calculation formula , calculate the task impact coefficient corresponding to each reference fault of each server ; Through the calculation formula , calculate the task impact coefficient corresponding to each reference fault of each storage device ; Through the calculation formula , calculate the task impact coefficient corresponding to each reference fault of each storage device , where , , , , , respectively represent the number of normally executed tasks and the number of abnormally executed tasks corresponding to each server, each storage device, and each network device at each fault time point, , , respectively represent the total number of execution tasks corresponding to each server, each storage device, and each network device at each failure time point. represents the number of each server. represents the number of each server corresponding to each failure. represents the number of each storage device. represents the number of each storage device corresponding to each failure. represents the number of each network device. represents the number of each network device corresponding to each failure; Through the failure time points of each server, each storage device, and each network device corresponding to each reference failure, statistical data is collected on the task resource requirements of each normal execution task and each abnormal execution task, obtaining the total task resource requirements of normal execution tasks and the total task resource requirements of abnormal execution tasks for each server, each storage device, and each network device corresponding to each reference failure, and obtaining the historical operation data corresponding to each failure time point of each server, each storage device, and each network device at each reference failure; Establish a data model for the total task resource requirements of normal execution tasks, the total task resource requirements of abnormal execution tasks, and the historical operation data of each server, each storage device, and each network device corresponding to each reference failure, and obtain the normal execution task resource occupancy ratio, abnormal execution task occupancy ratio, and total execution task occupancy ratio of each server, each storage device, and each network device corresponding to each reference failure through the data model; Through the calculation formula , calculate and obtain the comprehensive reference coefficient corresponding to each server for each reference failure ; Through the calculation formula , calculate and obtain the comprehensive reference coefficient corresponding to each storage device for each reference failure ; Through the calculation formula , calculate and obtain the comprehensive reference coefficient corresponding to each network device for each reference failure , where , , , , , respectively represent the normal execution task resource occupancy ratio and abnormal execution task occupancy ratio corresponding to each server, each storage device, and each network device at each failure time point. , , respectively represent the total occupancy ratios of the execution tasks corresponding to each server, each storage device, and each network device at each failure time point; Compare and analyze the comprehensive reference coefficients corresponding to each reference failure of each server, each storage device, and each network device with a preset comprehensive reference coefficient threshold. If the comprehensive reference coefficient is greater than the preset comprehensive reference coefficient, it indicates that the device failure is not caused by task allocation. If the comprehensive reference coefficient is less than or equal to the preset comprehensive reference coefficient, it indicates that the device failure is caused by task allocation; Statistically obtain the reference failures caused by task allocation and the reference failures not caused by task allocation corresponding to each server, each storage device, and each network device, and record the reference failures caused by task allocation and the reference failures not caused by task allocation corresponding to each server, each storage device, and each network device as the multi-modal data fusion result.

[0009] In a preferred embodiment of this solution, the specific execution manner of the optimization control module is as follows: Obtain the normal execution task resource occupancy ratio, abnormal execution task occupancy ratio, normal execution task number, and abnormal execution task number corresponding to each reference failure caused by task allocation of each server, each storage device, and each network device, and record the normal execution task resource occupancy ratio, abnormal execution task occupancy ratio, normal execution task number, and abnormal execution task number corresponding to each reference failure caused by task allocation as each task optimization allocation control template; Optimize and control the task allocation of the data center through each task optimization allocation control template.

[0010] To achieve the above object, the present invention also provides the following technical solution: An optimization method for a data center with multi-modal data fusion, including the following steps: Collect multi-type data corresponding to the data center to obtain multi-modal data corresponding to the data center; Process the multi-modal data corresponding to the data center to obtain multi-modal reference data corresponding to the data center; Perform data analysis based on the multi-modal reference data corresponding to the data center to obtain a multi-modal data fusion result corresponding to the data center; Perform optimization control based on the multi-modal data fusion result corresponding to the data center.

[0011] Compared with the prior art, the beneficial effects of the present invention are: By collecting various types of data corresponding to the data center, multi-modal data of the data center is obtained. Then, multi-modal parameter data is generated based on the multi-modal data. According to the multi-modal parameter data, combined with the fault data, task data, etc. in each server, each storage device, and each network device in the data, analysis is carried out to generate a multi-modal data fusion result. This multi-modal data fusion result records various problems that occur in the data center, and then, according to the various problems that occur, each task optimization allocation control template is generated, and optimization control is carried out according to this template. It improves the flexibility of the data center in optimizing task allocation and the efficiency of the data center in processing tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The present invention will be further described with reference to the accompanying drawings. However, the embodiments in the drawings do not constitute any limitation to the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to the following drawings.

[0013] Figure 1 It is a schematic diagram of module connection in an embodiment of the present invention.

[0014] Figure 2 It is a schematic diagram of step connection in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0015] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

[0016] Please refer to Figure 1 , the present invention provides an optimization system for a data center with multi-modal data fusion. The system includes a data collection module, a data preprocessing module, a data analysis module, and an optimization control module; The data collection module is connected to the data preprocessing module and the data analysis module. The data preprocessing module is connected to the data analysis module. The data analysis module is connected to the optimization control module.

[0017] The data collection module: is used to collect various types of data corresponding to the data center to obtain multi-modal data corresponding to the data center; Furthermore, the specific execution method of the data collection module is as follows: Obtain the historical operation data corresponding to each type of device through the data monitoring platform, where the device types include servers, storage devices, and network devices; The historical operation data of the server includes the CPU utilization rate, memory usage, disk I / O, and network bandwidth usage corresponding to each server; The historical operation data of the storage device includes the disk utilization rate, read / write rate, IOPS, and latency corresponding to each storage device; The historical operation data of the network device includes the bandwidth usage rate, packet loss rate, latency, throughput, and error rate corresponding to each network device; Obtain the historical task data of the data center through the task log. The historical task data includes each task executed by the data center, the task number of each task, the task basic data, the task execution time, the task termination time, the task resource requirements, the server number, storage device number, and network device number corresponding to each task; Among them, the task resource requirements include CPU requirements, network requirements, and storage requirements; Among them, the task basic data includes the computational complexity of the task and the data scale of the task; Record the historical operation data and historical task data corresponding to each type of device as the multimodal data corresponding to the data center.

[0018] Data preprocessing module: used to perform data processing on the multimodal data corresponding to the data center to obtain the multimodal reference data corresponding to the data center; Furthermore, the specific execution method of the data preprocessing module is as follows: Establish a data extraction relationship between the data preprocessing module and the database, and extract the standard task execution duration corresponding to each task execution model stored in the database. The task execution model includes the basic task data and task resource requirements, and obtain the standard task execution duration corresponding to each task through screening by the task basic data and task resource requirements corresponding to each task; Obtain the actual task execution duration corresponding to each task through the task execution time and task termination time corresponding to each task, compare the actual task execution duration corresponding to each task with the standard task execution duration. If the actual task execution duration corresponding to a certain task is less than or equal to the standard task execution duration, it means that the task is executed normally. If the actual task execution duration corresponding to a certain task is greater than the standard task execution duration, it means that the task is not executed normally, and count the normally executed tasks and the abnormally executed tasks; Obtain the numbers corresponding to each server, each storage device, and each network device in the data center, and perform matching and screening through the server number, storage device number, and network device number corresponding to each task to obtain the tasks corresponding to each server, each storage device, and each network device in the data center; Time filtering is performed based on the task execution time and task termination time of each task corresponding to each server, each storage device, and each network device in the data center, to obtain each normal execution task and each abnormal execution task corresponding to each server, each storage device, and each network device in each preset time period, and the total number of execution tasks, the number of normal execution tasks, and the number of abnormal execution tasks corresponding to each server, each storage device, and each network device in each preset time period are statistically obtained; The total number of execution tasks, the number of normal execution tasks, and the number of abnormal execution tasks corresponding to each server, each storage device, and each network device in each preset time period are recorded as the multimodal reference data corresponding to the data center.

[0019] Data analysis module: used to perform data analysis based on the multimodal reference data corresponding to the data center to obtain the multimodal data fusion result corresponding to the data center; Furthermore, the specific execution method of the data analysis module is as follows: Establish a data extraction relationship between the data analysis module and the database, and extract various fault types caused by resource allocation or task allocation stored in the database; Obtain the historical fault data corresponding to each numbered device through the fault and repair data log, where the historical fault data includes each fault corresponding to each numbered device, the fault type and fault time point corresponding to each fault. Match and screen the fault types corresponding to each fault of each numbered device with various fault types caused by resource allocation or task allocation, and record the fault types that match the fault types caused by resource allocation or task allocation as the reference fault types. Screen each fault corresponding to the reference fault type of each numbered device, which is recorded as each reference fault corresponding to each numbered device, and screen the fault degree and fault time point corresponding to each reference fault; Obtain each reference fault and fault time point corresponding to each server, each storage device, and each network device through the numbering corresponding to each server, each storage device, and each network device; Filter through the fault time points corresponding to each reference fault of each server, each storage device, and each network device to obtain each normal execution task, each abnormal execution task, the total number of execution tasks, the number of normal execution tasks, the number of abnormal execution tasks, and the historical operation data corresponding to each server, each storage device, and each network device at the fault time points corresponding to each reference fault; Through the calculation formula , calculate the task impact coefficient corresponding to each reference fault of each server ; Through the calculation formula , calculate the task impact coefficient corresponding to each reference fault of each storage device ; By using the calculation formula , the task impact coefficients corresponding to each storage device for each reference fault are calculated , where , , , , , respectively represent the number of normally executed tasks, the number of abnormally executed tasks corresponding to each server, each storage device, and each network device at each fault time point , , respectively represent the total number of executed tasks corresponding to each server, each storage device, and each network device at each fault time point represents the number of each server represents the number of each server corresponding to each fault represents the number of each storage device represents the number of each storage device corresponding to each fault represents the number of each network device represents the number of each network device corresponding to each fault; Based on the fault time points of each reference fault corresponding to each server, each storage device, and each network device, data statistics are performed on the task resource requirements of each normally executed task and each abnormally executed task, obtaining the total task resource requirements of the normally executed tasks and the total task resource requirements of the abnormally executed tasks corresponding to each server, each storage device, and each network device for each reference fault, and obtaining the historical operation data corresponding to each server, each storage device, and each network device at each fault time point of each reference fault; A data model is established for the total task resource requirements of the normally executed tasks, the total task resource requirements of the abnormally executed tasks, and the historical operation data corresponding to each server, each storage device, and each network device for each reference fault. Through the data model, the normal execution task resource occupancy ratio, the abnormal execution task occupancy ratio, and the total execution task occupancy ratio corresponding to each server, each storage device, and each network device for each reference fault are obtained; By using the calculation formula , the comprehensive reference coefficient corresponding to each server for each reference fault is calculated ; By using the calculation formula , the comprehensive reference coefficient corresponding to each storage device for each reference fault is calculated ; By using the calculation formula , the comprehensive reference coefficient corresponding to each network device for each reference fault is calculated , where , , , , , respectively represent the normal execution task resource occupancy ratio, abnormal execution task occupancy ratio corresponding to each server, each storage device, and each network device at each failure time point. , , respectively represent the total occupancy ratio of execution tasks corresponding to each server, each storage device, and each network device at each failure time point. Compare and analyze the comprehensive reference coefficient corresponding to each reference failure of each server, each storage device, and each network device with the preset comprehensive reference coefficient threshold. If the comprehensive reference coefficient is greater than the preset comprehensive reference coefficient, it indicates that the device failure is not caused by task allocation. If the comprehensive reference coefficient is less than or equal to the preset comprehensive reference coefficient, it indicates that the device failure is caused by task allocation. Statistically obtain the reference failures caused by task allocation and the reference failures not caused by task allocation corresponding to each server, each storage device, and each network device, and record the reference failures caused by task allocation and the reference failures not caused by task allocation corresponding to each server, each storage device, and each network device as the multi-modal data fusion result.

[0020] Optimization control module: used to perform optimization control according to the multi-modal data fusion result corresponding to the data center.

[0021] Furthermore, the specific execution method of the optimization control module is as follows: Obtain the normal execution task resource occupancy ratio, abnormal execution task occupancy ratio, normal execution task number, and abnormal execution task number corresponding to each reference failure caused by task allocation of each server, each storage device, and each network device, and record the normal execution task resource occupancy ratio, abnormal execution task occupancy ratio, normal execution task number, and abnormal execution task number corresponding to each reference failure caused by task allocation as each task optimization allocation control template. Optimize and control the task allocation of the data center through each task optimization allocation control template.

[0022] Please refer to Figure 2 , to achieve the above object, the present invention also provides the following technical solution: An optimization system for a data center with multi-modal data fusion, including the following steps: Collect various types of data corresponding to the data center to obtain multi-modal data corresponding to the data center; Process the multi-modal data corresponding to the data center to obtain multi-modal reference data corresponding to the data center; Perform data analysis based on the multimodal reference data corresponding to the data center to obtain the multimodal data fusion result corresponding to the data center; Perform optimization control based on the multimodal data fusion result corresponding to the data center.

[0023] The above are all preferred embodiments of this application. The protection scope of this application is not limited hereby. Therefore, all equivalent changes made according to the structure, shape, and principle of this application shall be covered within the protection scope of this application.

Claims

1. An optimized data center system for multi-modal data fusion, characterized in that: Including: Data acquisition module: used to acquire various types of data corresponding to the data center to obtain multimodal data corresponding to the data center; Data preprocessing module: used to perform data processing on the multimodal data corresponding to the data center to obtain multimodal reference data corresponding to the data center; Data analysis module: used to perform data analysis based on the multimodal reference data corresponding to the data center to obtain a multimodal data fusion result corresponding to the data center; Optimization control module: used to perform optimization control based on the multimodal data fusion result corresponding to the data center.

2. The optimized data center system for multimodal data fusion according to claim 1, characterized in that: The specific execution method of the data acquisition module is as follows: Obtain the historical operation data corresponding to each type of device through a data monitoring platform, where the device types include servers, storage devices, and network devices; The historical operation data of the server includes the CPU utilization rate, memory usage rate, disk I / O, and network bandwidth utilization rate corresponding to each server; The historical operation data of the storage device includes the disk utilization rate, read / write rate, IOPS, and latency corresponding to each storage device; The historical operation data of the network device includes the bandwidth utilization rate, packet loss rate, latency, throughput, and error rate corresponding to each network device; Obtain the historical task data of the data center through task logs, where the historical task data includes each task executed by the data center, the task number of each task, task basic data, task execution time, task termination time, task resource requirements, the server number, storage device number, and network device number corresponding to each task; Among them, the task resource requirements include CPU requirements, network requirements, and storage requirements; Among them, the task basic data includes the computational complexity and data scale of the task; Record the historical operation data and historical task data corresponding to each type of device as multimodal data corresponding to the data center.

3. The optimized data center system for multimodal data fusion according to claim 2, wherein: The specific execution method of the data preprocessing module is as follows: Establish a data extraction relationship between the data preprocessing module and the database, and extract the standard task execution duration corresponding to each task execution model stored in the database, where the task execution model includes task basic data and task resource requirements, and obtain the standard task execution duration corresponding to each task by filtering through the task basic data and task resource requirements corresponding to each task; Through the task execution time and task termination time corresponding to each task, obtain the actual task execution duration corresponding to each task, compare the actual task execution duration corresponding to each task with the standard task execution duration. If the actual task execution duration corresponding to a certain task is less than or equal to the standard task execution duration, it means that the task is executed normally. If the actual task execution duration corresponding to a certain task is greater than the standard task execution duration, it means that the task is executed abnormally. Statistically obtain each normally executed task and each abnormally executed task; Obtain the numbers corresponding to each server, each storage device, and each network device in the data center, and perform matching and screening through the server number, storage device number, and network device number corresponding to each task to obtain each task corresponding to each server, each storage device, and each network device in the data center; Time filtering is performed based on the task execution time and task termination time of each task corresponding to each server, each storage device, and each network device in the data center, to obtain each normal execution task and each abnormal execution task corresponding to each server, each storage device, and each network device in each preset time period, and the total number of execution tasks, the number of normal execution tasks, and the number of abnormal execution tasks corresponding to each server, each storage device, and each network device in each preset time period are statistically obtained; The total number of execution tasks, the number of normal execution tasks, and the number of abnormal execution tasks corresponding to each server, each storage device, and each network device in each preset time period are denoted as the multi-modal reference data corresponding to the data center.

4. The optimized data center system for multimodal data fusion according to claim 3, characterized in that: The specific execution method of the data analysis module is as follows: Establish a data extraction relationship between the data analysis module and the database, and extract various fault types caused by resource allocation or task allocation stored in the database; Obtain the historical fault data corresponding to each numbered device through the fault and repair data log. The historical fault data includes each fault corresponding to each numbered device, the fault type and fault time point corresponding to each fault. The fault types of each fault corresponding to each numbered device are matched and screened with various fault types caused by resource allocation or task allocation. The fault types that match the fault types caused by resource allocation or task allocation are denoted as reference fault types. The faults corresponding to the reference fault types for each numbered device are screened and denoted as the reference faults corresponding to each numbered device. The fault degree and fault time point corresponding to each reference fault are screened; The reference faults and fault time points corresponding to each server, each storage device, and each network device are obtained by screening according to the numbers corresponding to each server, each storage device, and each network device; Screening is performed based on the fault time points corresponding to each reference fault of each server, each storage device, and each network device, to obtain each normal execution task, each abnormal execution task, the total number of execution tasks, the number of normal execution tasks, the number of abnormal execution tasks, and the historical operation data corresponding to each server, each storage device, and each network device at the fault time points corresponding to each reference fault; By using the calculation formula , the task impact coefficients corresponding to each server for each reference fault are calculated ; By using the calculation formula , the task impact coefficients corresponding to each storage device for each reference fault are calculated ; By using the calculation formula , the task impact coefficients corresponding to each storage device for each reference fault are calculated , where , , , , , respectively represent the number of normally executed tasks, the number of abnormally executed tasks corresponding to each server, each storage device, and each network device at each fault time point , , respectively represent the total number of executed tasks corresponding to each server, each storage device, and each network device at each fault time point represents the number of each server represents the number of each server corresponding to each fault represents the number of each storage device represents the number of each storage device corresponding to each fault represents the number of each network device represents the number of each network device corresponding to each fault; Data statistics are performed on the task resource requirements of each normal execution task and each abnormal execution task corresponding to each server, each storage device, and each network device at the fault time points corresponding to each reference fault of each server, each storage device, and each network device, to obtain the total task resource requirements of the normal execution tasks and the total task resource requirements of the abnormal execution tasks corresponding to each server, each storage device, and each network device at each reference fault, and obtain the historical operation data corresponding to each server, each storage device, and each network device at each fault time point corresponding to each reference fault; A data model is established for the total task resource requirements of the normal execution tasks, the total task resource requirements of the abnormal execution tasks, and the historical operation data corresponding to each server, each storage device, and each network device at each reference fault. Through the data model, the resource occupancy ratio of the normal execution tasks, the occupancy ratio of the abnormal execution tasks, and the total occupancy ratio of the execution tasks corresponding to each server, each storage device, and each network device at each reference fault are obtained; By using the calculation formula , the comprehensive reference coefficients corresponding to each server for each reference fault are calculated ; By using the calculation formula , the comprehensive reference coefficients corresponding to each storage device for each reference fault are calculated ; By using the calculation formula , the comprehensive reference coefficients corresponding to each network device for each reference fault are calculated , where , , , , , respectively represent the resource occupancy ratios of normal task execution, the occupancy ratios of abnormal task execution corresponding to each server, each storage device, and each network device at each fault time point , , respectively represent the total occupancy ratios of task execution corresponding to each server, each storage device, and each network device at each fault time point; Compare the comprehensive reference coefficients corresponding to each server, each storage device, and each network device for each reference failure with the preset comprehensive reference coefficient threshold. If the comprehensive reference coefficient is greater than the preset comprehensive reference coefficient, it indicates that the device failure is not caused by task allocation. If the comprehensive reference coefficient is less than or equal to the preset comprehensive reference coefficient, it indicates that the device failure is caused by task allocation; Statistically obtain the reference failures caused by task allocation and the reference failures not caused by task allocation corresponding to each server, each storage device, and each network device, and record the reference failures caused by task allocation and the reference failures not caused by task allocation corresponding to each server, each storage device, and each network device as the multi-modal data fusion result.

5. The optimized data center system for multimodal data fusion according to claim 4, wherein: The specific execution method of the optimization control module is as follows: Obtain the normal execution task resource occupancy ratio, abnormal execution task occupancy ratio, normal execution task number, and abnormal execution task number corresponding to each reference failure caused by task allocation for each server, each storage device, and each network device, and record the normal execution task resource occupancy ratio, abnormal execution task occupancy ratio, normal execution task number, and abnormal execution task number corresponding to each reference failure caused by task allocation as each task optimization allocation control template; Optimize and control the task allocation of the data center through each task optimization allocation control template.

6. A method for optimizing a data center for multimodal data fusion, which is applied to the multimodal data fusion data center optimization system according to any one of claims 1-5, and is characterized in that: Including: Collect multi-type data corresponding to the data center to obtain multi-modal data corresponding to the data center; Process the multi-modal data corresponding to the data center to obtain multi-modal reference data corresponding to the data center; Perform data analysis based on the multi-modal reference data corresponding to the data center to obtain the multi-modal data fusion result corresponding to the data center; Perform optimization control based on the multi-modal data fusion result corresponding to the data center.

Images