A memory power consumption monitoring system for high performance computing clusters

Abstract

The present invention relates to the field of memory power consumption monitoring systems, and in particular to a high-performance computing cluster memory power consumption monitoring system, the system comprising a data acquisition module, which is used to acquire storage data of a task to be stored; a task allocation module, which is connected to the data acquisition module, and is used to allocate the task to be stored into a plurality of subtasks; a coordination degree detection module, which is connected to the task allocation module, and is used to determine the eligibility of the coordination degree of a calling process based on the computing nodes called by the plurality of subtasks; a disk analysis module, which is used to analyze the partition occupancy within a single partition, and comprises a disk storage unit for storing memory partitions of the plurality of subtasks, and an adjustment unit for adjusting memory fragments that cannot be stored in the disk storage unit; and a disk processing module, which is connected to the disk analysis module, and is used to process the memory fragments analyzed by the disk analysis module. The present invention improves the monitoring accuracy of the high-performance computing cluster memory power consumption monitoring system.

Description

Technical Field

[0001] The present invention relates to the technical field of memory power consumption monitoring systems, and in particular to a high-performance computing cluster memory power consumption monitoring system. Background Art

[0002] Currently, high-performance computing clusters play a key role in large-scale data processing and complex computing tasks. However, with the continuous increase in computing tasks, cluster memory power consumption monitoring has become an important challenge. Traditional power consumption monitoring methods often rely on manual detection or simple monitoring equipment, which have problems such as low monitoring accuracy and weak real-time performance, and cannot meet the needs of power consumption monitoring during cluster operation. To solve this problem, it is necessary to develop a high-performance computing cluster memory power consumption monitoring system to improve monitoring accuracy.

[0003] Chinese Patent Publication No.: CN105607726A. A method and device for reducing the memory power consumption of a high-performance computing cluster is disclosed, the method comprising: real-time monitoring of the operation status of the high-performance computing cluster; analyzing the types of jobs running in the high-performance computing cluster to determine the memory fault tolerance mechanism of the jobs running in the high-performance computing cluster; obtaining the fault tolerance level of the memory fault tolerance mechanism, and allocating the memory power consumption of the high-performance computing cluster according to the fault tolerance level. However, there is a problem in the prior art that the monitoring accuracy of the memory power consumption monitoring system of the high-performance computing cluster is low. Summary of the invention

[0004] To this end, the present invention provides a high-performance computing cluster memory power consumption monitoring system to overcome the problem of low accuracy of the high-performance computing cluster memory power consumption monitoring system in the prior art.

[0005] To achieve the above object, the present invention provides a high performance computing cluster memory power consumption monitoring system, comprising:

[0006] A data acquisition module, which is used to acquire storage data of the task to be stored;

[0007] A task allocation module, connected to the data acquisition module, for allocating the task to be stored into a plurality of subtasks;

[0008] A coordination degree detection module, which is connected to the task allocation module and is used to determine the eligibility of the coordination degree of the calling process based on the computing nodes called by a number of subtasks;

[0009] a disk analysis module, which is used to analyze the partition occupancy in a single partition, and includes a disk storage unit for storing the memory partitions of the plurality of subtasks, and an adjustment unit for adjusting the memory segments that cannot be stored in the disk storage unit;

[0010] The disk processing module is connected to the disk analysis module and is used to process the memory segments analyzed by the disk analysis module.

[0011] Furthermore, the coordination degree detection module is used to determine the eligibility of the coordination degree of the calling process by comparing the coordination degree evaluation values ​​of several computing nodes with the preset coordination degree evaluation values ​​of computing nodes;

[0012] If the cooperation degree evaluation values ​​of the plurality of computing nodes are greater than the preset computing node cooperation degree evaluation value, it is determined that the cooperation degree of the calling process is qualified;

[0013] If the cooperation degree evaluation values ​​of the plurality of computing nodes are less than the preset computing node cooperation degree evaluation value, it is determined that the cooperation degree of the calling process is unqualified.

[0014] The disk analysis module analyzes the target disk to determine whether the target disk has a disk partition matching a single subtask under the condition that the coordination degree of the plurality of computing nodes is unqualified.

[0015] If the memory of the disk partition is larger than a preset ratio of the data volume of the single subtask, the disk analysis module determines that the disk partition matches the single subtask and stores the storage data corresponding to the single subtask.

[0016] Furthermore, the disk analysis module determines the order in which the target disk executes storage, and determines whether there is a preceding disk partition before the location of the disk partition in the target disk and the remaining memory of the preceding disk partition is greater than the data volume of the single subtask;

[0017] If the remaining memory of the front disk partition is greater than the data volume of the single subtask, the front disk partition is processed;

[0018] If there is no front disk partition with remaining memory larger than the data volume of the single subtask, the storage task of the single subtask is directly executed.

[0019] Furthermore, the disk analysis module determines whether there is pre-storage data in the pre-storage disk partition under the condition that the pre-storage disk partition is processed. If there is no pre-storage data, the corresponding pre-storage disk partition is hidden. If there is pre-storage data, the importance of the pre-storage data is analyzed.

[0020] Furthermore, the disk analysis module determines a processing method for processing the front disk partition according to a comparison result between the importance evaluation value and a preset importance evaluation value;

[0021] If the importance evaluation value is less than the preset importance evaluation value, the disk processing module determines to hide the front storage data of the front disk partition;

[0022] If the importance evaluation value is greater than the preset importance evaluation value, the disk processing module determines that the front storage data of the front disk partition is to be overwritten.

[0023] Furthermore, under the condition that the disk storage unit directly executes the storage task of the single subtask, it determines whether the data volume of the single subtask can be directly placed into the target disk partition. If the data volume of the single subtask can be directly placed into the front disk partition, it is stored in the front disk partition; if the data volume of the single subtask cannot be directly placed into the front disk partition, a historical accuracy analysis is performed on the front disk partition.

[0024] Furthermore, the disk analysis module determines whether to adjust the preset coordination evaluation value according to a comparison result between the historical accuracy rate and the preset historical accuracy rate under the condition that the data volume of the single subtask cannot be directly put into the front disk partition;

[0025] If the historical accuracy rate of the front disk partition that cannot directly fit the data volume of the single subtask is greater than or equal to the preset historical accuracy rate, then determining to adjust the preset coordination degree;

[0026] If the historical accuracy of the front disk partition into which the data volume of the single subtask cannot be directly placed is less than the preset historical accuracy, it is determined not to adjust the preset coordination degree.

[0027] Further, the disk analysis module determines the storage fluency evaluation value of the single subtask under the condition of determining the storage task execution to determine whether the storage of the single subtask meets the standard;

[0028] If the storage fluency evaluation value of the single subtask is less than the preset storage fluency evaluation value, it is determined that the storage of the single subtask does not meet the standard, and the storage node is switched to select the next disk partition;

[0029] If the storage fluency evaluation value of the single subtask is greater than or equal to a preset storage fluency evaluation value, it is determined that the storage of the single subtask meets the standard.

[0030] Further, the disk analysis module determines whether the stability of the single subtask in the storage node meets the standard according to the comparison result of the historical data loss rate and the preset historical data loss rate, under the condition that the storage of the single subtask does not meet the standard and the storage node is switched;

[0031] If the historical data loss rate is less than or equal to the preset historical data loss rate, it is determined that the stability of the single subtask on the storage node meets the standard;

[0032] If the historical data loss rate is greater than the preset historical data loss rate, it is determined that the stability of the single subtask in the storage node does not meet the standard.

[0033] Further, when the disk analysis unit determines that the stability of the single subtask on the storage node does not meet the standard, the adjustment unit determines the adjustment method according to a comparison result between the difference between the historical data loss rate and the preset historical data loss rate and the preset difference;

[0034] If the difference is less than or equal to the preset difference, determining to increase the data volume of the single subtask by a preset proportion;

[0035] If the difference is greater than the preset difference, it is determined to increase the preset matching degree evaluation value.

[0036] Compared with the prior art, the beneficial effect of the present invention is that the present invention determines the eligibility of the computing node in the calling process by comparing the computing node's coordination evaluation value with the preset computing node coordination evaluation value through the coordination degree detection module. By evaluating the coordination degree of the computing nodes, the collaboration efficiency between the computing nodes is improved.

[0037] Furthermore, the coordination degree detection module compares the coordination degree evaluation value of the computing node with the preset coordination degree evaluation value of the computing node to determine the eligibility of the computing node in the calling process. By evaluating the coordination degree of the computing nodes, the collaboration efficiency between the computing nodes and the accuracy of the monitoring system are improved.

[0038] Furthermore, for the amount of data that cannot be directly stored in a single subtask of the target disk partition, a historical accuracy analysis is performed, and a preset coordination evaluation value is added based on the difference between the preset historical data accuracy and the historical data accuracy, and the storage fluency evaluation value is calculated to determine whether the storage task meets the standard. If the storage task does not meet the standard, the storage node is switched, and if the historical data loss rate is greater than the preset historical data loss rate, it is determined that the stability of the single subtask in the storage node does not meet the standard. Determining stability helps to improve the stability and accuracy of the monitoring system.

[0039] Furthermore, the disk analysis unit is adjusted in the case of stability failure, and determines the preset proportion of the data volume of a single subtask or the preset coordination evaluation value according to the difference between the historical data loss rate and the preset historical data loss rate. Through adjustment, the system can effectively improve the monitoring accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a schematic diagram of module connections of a high performance computing cluster memory power consumption monitoring system according to an embodiment of the present invention;

[0041] Figure 2 A flow chart for determining the eligibility of the coordination degree of a calling process according to an embodiment of the present invention;

[0042] Figure 3 A flowchart of a method for processing a front disk partition according to an embodiment of the present invention;

[0043] Figure 4 The present invention is a flowchart of adjusting a preset cooperation degree evaluation value according to an embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0045] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0046] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is merely for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0047] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0048] The high performance computing cluster memory power consumption monitoring system according to the embodiment of the present invention comprises:

[0049] A data acquisition module, which is used to acquire storage data of the task to be stored;

[0050] A task allocation module, connected to the data acquisition module, for allocating the task to be stored into a plurality of subtasks;

[0051] A coordination degree detection module, which is connected to the task allocation module and is used to determine the eligibility of the coordination degree of the calling process based on the computing nodes called by a number of subtasks;

[0052] a disk analysis module, which is used to analyze the partition occupancy in a single partition, and includes a disk storage unit for storing the memory partitions of the plurality of subtasks, and an adjustment unit for adjusting the memory segments that cannot be stored in the disk storage unit;

[0053] The disk processing module is connected to the disk analysis module and is used to process the memory segments analyzed by the disk analysis module.

[0054] Specifically, the coordination degree detection module is used to determine the eligibility of the coordination degree of the calling process by comparing the coordination degree evaluation values ​​of several computing nodes with the preset computing node coordination degree evaluation value 0.9;

[0055] If the cooperation degree evaluation values ​​of the plurality of computing nodes are greater than the preset computing node cooperation degree evaluation value, it is determined that the cooperation degree of the calling process is qualified;

[0056] If the cooperation degree evaluation values ​​of the plurality of computing nodes are less than the preset computing node cooperation degree evaluation value, it is determined that the cooperation degree of the calling process is unqualified.

[0057] Specifically, the matching degree detection module calculates the matching degree evaluation value according to the following formula, setting:

[0058] W=T / Tz+Y / Yz

[0059] Wherein, W represents the cooperation degree evaluation value, T represents the number of called computing nodes, Tz represents the number of remaining uncalled computing nodes, Y represents the number of subtasks of called computing nodes, and Yz represents the number of subtasks of uncalled computing nodes.

[0060] Specifically, the disk analysis module analyzes the target disk to determine whether the target disk has a disk partition matching a single subtask under the condition that the cooperation degree of the plurality of computing nodes is unqualified.

[0061] If the memory of the disk partition is larger than 80% of the preset ratio of the data volume of the single subtask, the disk analysis module determines that the disk partition matches the single subtask and stores the storage data corresponding to the single subtask.

[0062] Specifically, the embodiment of the present invention determines the eligibility of a computing node during the calling process by comparing the computing node's coordination evaluation value with a preset computing node coordination evaluation value through a coordination degree detection module. By evaluating the coordination degree of computing nodes, the collaboration efficiency between computing nodes is improved.

[0063] Specifically, the disk analysis module determines the order in which the target disk executes storage, and determines whether there is a preceding disk partition before the location of the disk partition in the target disk and the remaining memory of the preceding disk partition is greater than the data volume of the single subtask;

[0064] If the remaining memory of the front disk partition is greater than the data volume of the single subtask, the front disk partition is processed;

[0065] If there is no front disk partition with remaining memory equal to the data volume of the single subtask, the storage task of the single subtask is directly executed.

[0066] Specifically, the disk analysis module determines whether there is pre-storage data in the pre-storage disk partition under the condition that the pre-storage disk partition is processed. If there is no pre-storage data, the corresponding pre-storage disk partition is hidden. If there is pre-storage data, the importance of the pre-storage data is analyzed.

[0067] Specifically, the disk analysis module characterizes the importance of the pre-stored data according to an importance evaluation value, and the importance evaluation value is calculated according to the following formula and is set:

[0068]

[0069] Among them, Q represents the importance evaluation value, S represents the proportion of encryption modules in the disk partition, W represents the proportion of sensitive modules in the disk partition, and n represents the number of modules into which the disk partition is divided.

[0070] Specifically, the disk analysis module determines the processing method for processing the front disk partition according to the comparison result between the importance evaluation value and the preset importance evaluation value 0.9;

[0071] If the importance evaluation value is less than the preset importance evaluation value, the disk processing module determines to hide the front storage data of the front disk partition;

[0072] If the importance evaluation value is greater than the preset importance evaluation value, the disk processing module determines to overwrite the pre-stored data of the pre-disk partition.

[0073] Specifically, under the condition that the disk storage unit directly executes the storage task of the single subtask, it determines whether the data volume of the single subtask can be directly placed into the target disk partition. If the data volume of the single subtask can be directly placed into the front disk partition, it is stored in the front disk partition; if the data volume of the single subtask cannot be directly placed into the front disk partition, a historical accuracy analysis is performed on the front disk partition.

[0074] Specifically, the disk analysis module determines whether to adjust the preset coordination evaluation value according to the comparison result between the historical accuracy rate and the preset historical accuracy rate of 85% under the condition that the data volume of the single subtask cannot be directly put into the front disk partition;

[0075] If the historical accuracy rate of the front disk partition that cannot directly fit the data volume of the single subtask is greater than or equal to the preset historical accuracy rate, then determining to adjust the preset coordination degree;

[0076] If the historical accuracy of the front disk partition into which the data volume of the single subtask cannot be directly placed is less than the preset historical accuracy, it is determined not to adjust the preset coordination degree.

[0077] Specifically, the adjustment unit increases the preset matching degree evaluation value according to the comparison result of the preset historical data accuracy rate and the difference between the historical data accuracy rate and the preset difference value of 3.5%;

[0078] If the difference is less than or equal to the preset difference, the preset matching degree evaluation value is increased to a corresponding value by a first preset increase adjustment coefficient of 1.1;

[0079] If the difference is greater than the preset difference, the preset matching degree evaluation value is increased to a corresponding value using a second preset increase adjustment coefficient of 1.15.

[0080] The difference is the difference between the preset historical data accuracy and the historical data accuracy.

[0081] In an embodiment of the present invention, the increased degree of fit evaluation value is set to Lc, and Lc=L×Ki is set, wherein L represents the preset degree of fit evaluation value, Ki represents the i-th preset increase adjustment coefficient, i takes a value of 1 or 2, K1 is the first preset increase adjustment coefficient, and K2 is the second preset increase adjustment coefficient.

[0082] Specifically, the fluency evaluation value is calculated according to the following formula:

[0083] P=k / t

[0084] Wherein, P represents the fluency evaluation value, k represents the amount of stored data, and t represents the time used to store the data.

[0085] Specifically, the disk analysis module determines whether the storage of the single subtask meets the storage standard by comparing the storage fluency evaluation value of the single subtask with the preset storage fluency evaluation value 0.8 under the condition of determining the execution of the storage task;

[0086] If the storage fluency evaluation value of the single subtask is less than the preset storage fluency evaluation value, it is determined that the storage of the single subtask does not meet the standard, and the storage node is switched to select the next disk partition;

[0087] If the storage fluency evaluation value of the single subtask is greater than or equal to a preset storage fluency evaluation value, it is determined that the storage of the single subtask meets the standard.

[0088] Specifically, the disk analysis module determines whether the stability of the single subtask in the storage node meets the standard based on the comparison result of the historical data loss rate and the preset historical data loss rate of 5%, under the condition that the storage of the single subtask does not meet the standard and the storage node is switched;

[0089] If the historical data loss rate is less than or equal to the preset historical data loss rate, it is determined that the stability of the single subtask on the storage node meets the standard;

[0090] If the historical data loss rate is greater than the preset historical data loss rate, it is determined that the stability of the single subtask in the storage node does not meet the standard.

[0091] The embodiment of the present invention performs a historical accuracy analysis on the amount of data that cannot be directly stored in a single subtask of the target disk partition, increases a preset coordination evaluation value based on the difference between the preset historical data accuracy and the historical data accuracy, and determines whether the storage task meets the standard based on the calculated storage fluency evaluation value. If the storage task does not meet the standard, the storage node is switched, and if the historical data loss rate is greater than the preset historical data loss rate, it is determined that the stability of the single subtask in the storage node does not meet the standard. Determining stability helps to improve the stability and accuracy of the monitoring system.

[0092] Specifically, when the disk analysis unit determines that the stability of the single subtask on the storage node does not meet the standard, the adjustment unit determines the adjustment method according to the comparison result between the difference between the historical data loss rate and the preset historical data loss rate and the preset difference of 2%;

[0093] If the difference is less than or equal to the preset difference, determining to increase the data volume of the single subtask by a preset proportion;

[0094] If the difference is greater than the preset difference, it is determined to increase the preset matching degree evaluation value.

[0095] The difference is the difference between the historical data loss rate and the preset historical data loss rate.

[0096] Specifically, the adjustment unit determines a preset proportion for increasing the data amount of the single subtask according to the difference between the preset difference and the difference value of 1%;

[0097] If the difference between the preset difference and the difference is smaller than the preset difference, the preset proportion of the data volume of the single subtask is increased to a corresponding value by a first preset adjustment coefficient of 1.02;

[0098] If the difference between the preset difference and the difference is greater than the preset difference, the preset proportion of the data volume of the single subtask is increased to a corresponding value by a second preset adjustment coefficient of 1.05;

[0099] In an embodiment of the present invention, the increased preset ratio is set to Nc, and Nc=N×Ki is set, where N represents the preset ratio of the data volume of a single subtask, Ki represents the i-th preset adjustment coefficient, i takes a value of 1 or 2, K1 is the first preset adjustment coefficient, and K2 is the second preset adjustment coefficient.

[0100] Specifically, the adjustment unit determines to increase the preset matching degree evaluation value according to the difference between the difference and the preset difference of 1.5%;

[0101] If the difference is less than or equal to the preset difference, the matching degree evaluation value is increased to a corresponding value by using a first preset increase adjustment coefficient of 1.03;

[0102] If the difference is greater than the preset difference, the matching degree evaluation value is increased to a corresponding value using a second preset increase adjustment coefficient of 1.07.

[0103] In an embodiment of the present invention, the increased coordination degree evaluation value is set to Pz, and Pz=P×Ki is set, wherein P represents the coordination degree of a number of nodes, Ki represents the i-th preset increase adjustment coefficient, i takes a value of 1 or 2, K1 is the first preset increase adjustment coefficient, and K2 is the second preset increase adjustment coefficient.

[0104] In the embodiment of the present invention, the disk analysis unit is adjusted when the stability does not meet the standard, and the preset proportion of the data volume of a single subtask or the preset coordination evaluation value is increased according to the difference between the historical data loss rate and the preset historical data loss rate. Through adjustment, the system can effectively improve the monitoring accuracy.

[0105] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0106] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A high performance computing cluster memory power consumption monitoring system, characterized in that: include: A data acquisition module, which is used to acquire storage data of the task to be stored; A task allocation module, connected to the data acquisition module, for allocating the task to be stored into a plurality of subtasks; A coordination degree detection module, which is connected to the task allocation module, is used to determine the eligibility of the coordination degree of the calling process based on the computing nodes called by the subtasks, and if the coordination degree evaluation values ​​of the computing nodes are less than the preset computing node coordination degree evaluation values, determine that the coordination degree of the calling process is unqualified; The matching degree detection module calculates the matching degree evaluation value according to the following formula, setting: W=T / Tz+Y / Yz Wherein, W represents the cooperation degree evaluation value, T represents the number of computing nodes that have been called, Tz represents the number of computing nodes that have not been called, Y represents the number of subtasks of the computing nodes that have been called, and Yz represents the number of subtasks of the computing nodes that have not been called; a disk analysis module, which is used to analyze the partition occupancy in a single partition, and includes a disk storage unit for storing the memory partitions of the plurality of subtasks, and an adjustment unit for adjusting the memory segments that cannot be stored in the disk storage unit; a disk processing module connected to the disk analysis module to process the memory fragments analyzed by the disk analysis module, and to determine to hide the pre-stored data of the pre-stored disk partition if the importance evaluation value is less than a preset importance evaluation value, and to overwrite the pre-stored data of the pre-stored disk partition if the importance evaluation value is greater than the preset importance evaluation value; The disk analysis module characterizes the importance of the pre-stored data according to the importance evaluation value, and the importance evaluation value is calculated according to the following formula and is set: Among them, Q represents the importance evaluation value, S represents the proportion of encryption modules in the disk partition, W represents the proportion of sensitive modules in the disk partition, and n represents the number of modules into which the disk partition is divided.

2. The high performance computing cluster memory power consumption monitoring system according to claim 1, characterized in that: The coordination degree detection module determines the eligibility of the computing node during the calling process by comparing the coordination degree evaluation value of the computing node with the preset computing node coordination degree evaluation value. If the coordination degree is unqualified, the disk analysis module analyzes the target disk to confirm whether there is a disk partition that matches the single subtask.

3. The high performance computing cluster memory power consumption monitoring system according to claim 2, characterized in that: The disk analysis module determines the order in which the target disk executes storage, and determines whether there is a preceding disk partition before the location of the disk partition in the target disk and the remaining memory of the preceding disk partition is greater than the data volume of the single subtask, and determines that there is a preceding disk partition with a remaining memory greater than the data volume of the single subtask, processes the preceding disk partition, determines that there is no preceding disk partition with a remaining memory greater than the data volume of the single subtask, and directly executes the storage task of the single subtask.

4. The high performance computing cluster memory power consumption monitoring system according to claim 3, characterized in that: The disk analysis module processes the front disk partition to determine whether there is front storage data in the front disk partition. If there is no front storage data, the corresponding front disk partition is hidden. If there is front storage data, the importance of the front storage data is analyzed.

5. The high performance computing cluster memory power consumption monitoring system according to claim 4, characterized in that: The disk analysis module determines a processing method for the front disk partition based on a comparison result between the importance evaluation value and a preset importance evaluation value. If the importance evaluation value is less than the preset importance evaluation value, the disk processing module determines to hide the front storage data of the front disk partition. If the importance evaluation value is greater than the preset importance evaluation value, the disk processing module determines to overwrite the front storage data of the front disk partition.

6. The high performance computing cluster memory power consumption monitoring system according to claim 5, characterized in that: Under the condition that the disk storage unit directly executes the storage task of the single subtask, it is determined whether the data volume of the single subtask can be directly placed into the target disk partition.

7. The high performance computing cluster memory power consumption monitoring system according to claim 6, characterized in that: The disk analysis module determines whether to adjust the preset coordination degree evaluation value according to the comparison result between the historical accuracy rate and the preset historical accuracy rate under the condition that the data volume of the single subtask cannot be directly put into the front disk partition.

8. The high performance computing cluster memory power consumption monitoring system according to claim 7, characterized in that: The disk analysis module determines the storage fluency evaluation value of the single subtask under the condition of determining the storage task execution to determine whether the storage of the single subtask meets the standard.

9. The high performance computing cluster memory power consumption monitoring system according to claim 8, characterized in that: The disk analysis module determines whether the stability of the single subtask in the storage node meets the standard based on the comparison result of the historical data loss rate and the preset historical data loss rate, under the condition that the storage of the single subtask does not meet the standard and the storage node is switched.

10. The high performance computing cluster memory power consumption monitoring system according to claim 9, characterized in that: When the disk analysis unit determines that the stability of the single subtask on the storage node does not meet the standard, the adjustment unit determines whether to increase the data volume of the single subtask by a preset proportion or to increase the preset coordination degree evaluation value based on a comparison result of the difference between the historical data loss rate and the preset historical data loss rate and the preset difference.

Images