A distributed high performance computing architecture and computing method thereof

Abstract

The present invention discloses a distributed high-performance computing architecture and a computing method thereof. The distributed high-performance computing architecture includes a main center, a secondary center and a tertiary center. The main center is configured to globally manage computing power and formulate a task allocation strategy. The secondary center is configured to be a bridge between the main center and the tertiary center, which aggregates computing power and forwards tasks. The tertiary center is configured to allocate the data forwarded by the secondary center to each device node and report computing power. The high-performance distributed computing system provided by the present invention is easy to deploy the system, convenient to manage equipment and easy to manage resources. Decentralization allows the centers at all levels to be highly coupled, which not only reduces the load of the main center, but also makes it easier to maintain the system. The present invention effectively improves the stability of the system through decentralization. When a problem occurs in the upper center, the task can be directly sent to the lower center to perform the task, thereby ensuring the normal operation of the entire system.

Description

Technical Field

[0001] The present invention relates to the technical field of distributed computing in power grids, and in particular to a distributed high-performance computing architecture and a computing method thereof. Background Art

[0002] With the construction of my country's ultra-high voltage interconnected large power grid and the access of large-scale distributed power sources, in order to support the dispatching business of the power system, it is objectively required that the large power grid dispatching control system have a larger grid modeling range and higher accuracy. The grid voltage level has developed upward to the 1000kV ultra-high voltage grid and extended downward to the refined modeling of distributed power sources accessing the grid, and the scale of the grid is expanding rapidly. In order to meet the needs of cross-regional and cross-voltage level collaborative computing and ultra-large-scale grid computing in large-scale interconnected power grids, the new generation of smart grid dispatching and control systems proposes an overall architecture for distributed analysis and computing of large power grids, and the grid model is the basis of power system analysis and computing. In this regard, it is necessary to develop an instant sharing method of grid models suitable for online distributed computing of large power grids, and provide a solution to the problem of sharing online distributed computing models of large power grids under the multi-level collaborative dispatching mode.

[0003] However, existing distributed computing systems have shortcomings in computing power management and task allocation, such as the inability to accurately grasp the computing power of each node in real time, and unreasonable task allocation leading to low resource utilization. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] Therefore, the purpose of the present invention is to provide a distributed high-performance computing architecture and its computing method, which adopts a decentralized approach to evenly distribute tasks, reduce the overall center load, and avoid the instability of one link in the computing system affecting the operation of the entire computing system. At the same time, compared with traditional architectures, its resource recovery and reallocation have significant advantages.

[0006] To solve the above technical problems, according to one aspect of the present invention, the present invention provides the following technical solutions:

[0007] A distributed high-performance computing architecture includes a main center, a secondary center, and a tertiary center;

[0008] The central center is configured to: globally manage computing power and formulate task allocation strategies;

[0009] The secondary center is configured as: a bridge between the main center and the tertiary center, aggregating computing capabilities and forwarding tasks;

[0010] The third-level center is configured to distribute the information forwarded by the second-level center to each device node and report the computing capacity.

[0011] As a preferred solution of the distributed high-performance computing architecture described in the present invention, the central center is configured to: globally manage the computing power and formulate the task allocation strategy as follows:

[0012] The third-level center reports its computing capacity to the second-level center, and the second-level center reports its overall computing capacity to the main center;

[0013] The main center assigns tasks to each secondary center based on the computing capacity reported by each secondary center.

[0014] As a preferred solution of a distributed high-performance computing architecture described in the present invention, the secondary center is configured as a bridge between the main center and the tertiary center, and the aggregation of computing power and forwarding of tasks are specifically as follows:

[0015] The secondary center assigns tasks to the tertiary center according to its computing power and issues them down layer by layer; the lower-level center feeds back the computing results to the upper-level center, and finally the secondary center feeds back the computing results to the main center, which obtains the computing results through the main center.

[0016] As a preferred solution of the distributed high-performance computing architecture described in the present invention, a computing capacity evaluation module is deployed in the third-level center and the second-level center, and the computing capacity evaluation module first collects CPU usage, memory idleness and network bandwidth data in each center at regular intervals;

[0017] Secondly, analyze the historical data of CPU utilization to determine whether the system has insufficient CPU resources; analyze the free memory to determine memory leaks or over-allocation; analyze the network bandwidth data to determine the bandwidth usage status;

[0018] Then, based on the analysis results of the performance data, the CPU usage, memory idleness and network bandwidth are optimized, and the computing power is reported to the superior center.

[0019] As a preferred solution of a distributed high-performance computing architecture described in the present invention, the main center and the secondary center are also configured with a task recovery module, and the task recovery module is used to execute: when a problem occurs in the subordinate center, the main center and the secondary center will recover and reset the task, so as to ensure the normal operation of the entire system.

[0020] A computing method for a distributed high-performance computing architecture, the specific steps are as follows:

[0021] S1. The third-level center reports its computing capacity to the second-level center, and the second-level center reports its overall computing capacity to the main center;

[0022] S2. The main center assigns tasks to each secondary center according to the computing capacity reported by each secondary center. The secondary center assigns tasks to the tertiary center according to its computing capacity, and issues tasks layer by layer.

[0023] S3, the third-level center feeds back the calculation results to the second-level center, and finally the second-level center feeds back the calculation results to the main center, and the calculation results are obtained through the main center;

[0024] S4. During the sending process, when a secondary center or a tertiary center is offline, the sent task data is directly copied to the center, so that the center can assign tasks to each device node.

[0025] Compared with the prior art, the present invention has the following beneficial effects:

[0026] 1. The high-performance distributed computing system provided by the present invention is easy to deploy, manage equipment and manage resources. Decentralization allows centers at all levels to be highly coupled, which not only reduces the total center load, but also makes system maintenance easier.

[0027] 2. The present invention effectively improves the system stability through a decentralized approach. When a problem occurs in the upper-level center, the task can be directly sent to the lower-level center to execute the task; when a problem occurs in the lower-level center, the upper-level center can recover and reset the task, thereby ensuring the normal operation of the entire system. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the present invention will be described in detail below in combination with the accompanying drawings and detailed embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0029] Figure 1 It is a schematic diagram of the online mode of the distributed high performance computing architecture of the present invention;

[0030] Figure 2 This is a schematic diagram of the offline mode of the distributed high performance computing architecture of the present invention. DETAILED DESCRIPTION

[0031] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0032] Secondly, the present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention in detail, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0033] To make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0034] The present invention provides a distributed high-performance computing architecture and a computing method thereof, which adopts a decentralized approach, can evenly distribute tasks, reduce the overall center load, and avoid the instability of one link of the computing system affecting the operation of the entire computing system. At the same time, compared with the traditional architecture, its resource recovery and reallocation have significant advantages.

[0035] A distributed high performance computing architecture, comprising a main center 100, a secondary center 200 and a tertiary center 300;

[0036] The central center 100 is configured to: globally manage computing power and formulate task allocation strategies. Specifically, the third-level center 300 reports computing power to the second-level center 200, and the second-level center 200 reports the overall computing power of the center to the central center 100. The central center 100 allocates tasks to each second-level center 200 according to the computing power reported by each second-level center 200.

[0037] The secondary center 200 is configured as: a bridge between the main center 100 and the tertiary center 300, which aggregates computing power and forwards tasks. Specifically, the secondary center assigns tasks to the tertiary center according to its computing power and issues them layer by layer; the lower-level center feeds back the computing results to the upper-level center, and finally the secondary center feeds back the computing results to the main center, and the computing results are obtained through the main center.

[0038] The third-level center 300 is configured to distribute the information forwarded by the second-level center 200 to each device node and report the computing capacity.

[0039] Among them, a computing capacity evaluation module is deployed in the third-level center 300 and the second-level center 200. The computing capacity evaluation module first regularly collects the CPU usage, memory idle amount and network bandwidth data in each center, and then analyzes the historical data of CPU utilization to determine whether the system has a problem of insufficient CPU resources; analyzes the memory idle amount to determine memory leaks or over-allocation, analyzes the network bandwidth data to determine the broadband occupancy status, and then optimizes the CPU usage, memory idle amount and network bandwidth according to the analysis results of the performance data, and reports the computing capacity to the superior center.

[0040] The main center 100 and the secondary center 200 are also equipped with a task recovery module, which is used to execute: when a problem occurs in the subordinate center, the main center 100 and the secondary center 200 will recover and reset the task, thereby ensuring the normal operation of the entire system.

[0041] like Figure 1 As shown, the calculation method of the distributed high-performance computing architecture has the following specific steps:

[0042] S1. The third-level center 300 reports its computing capacity to the second-level center 200, and the second-level center 200 reports its overall computing capacity to the main center 100;

[0043] S2. The main center 100 allocates tasks to each secondary center 200 according to the computing capacity reported by each secondary center 200. The secondary center 200 allocates tasks to the tertiary center 300 according to its computing capacity, and issues tasks layer by layer.

[0044] S3. The third-level center 300 feeds back the calculation result to the second-level center 200. Finally, the second-level center 200 feeds back the calculation result to the main center 100, and the calculation result is obtained through the main center 100.

[0045] like Figure 2 As shown, during the sending process, when a secondary center 200 or a tertiary center 300 is offline, the sent task data is directly copied to the center, so that the center can assign tasks to each device node.

[0046] Although the present invention has been described above with reference to the embodiments, various modifications may be made thereto and parts thereof may be replaced by equivalents without departing from the scope of the present invention. In particular, as long as there is no structural conflict, the various features in the embodiments disclosed in the present invention may be used in combination with each other in any manner, and the fact that these combinations are not exhaustively described in this specification is only for the sake of omitting space and saving resources. Therefore, the present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A distributed high performance computing architecture, characterized in that: It includes the main center (100), the secondary center (200) and the tertiary center (300); The central center (100) is configured to: globally manage computing power and formulate task allocation strategies; The secondary center (200) is configured as: a bridge between the main center (100) and the tertiary center (300), aggregating computing capabilities and forwarding tasks; The third-level center (300) is configured to distribute the information forwarded by the second-level center (200) to each device node and report the computing capacity.

2. A distributed high performance computing architecture according to claim 1, characterized in that: The central center (100) is configured to: globally manage computing power and formulate task allocation strategies as follows: The third-level center (300) reports its computing capacity to the second-level center (200), and the second-level center (200) reports its overall computing capacity to the main center (100); The main center (100) allocates tasks to each secondary center (200) according to the computing capacity reported by each secondary center (200).

3. A distributed high performance computing architecture according to claim 1, characterized in that: The secondary center (200) is configured as a bridge between the main center (100) and the tertiary center (300), and the aggregation of computing power and forwarding of tasks are as follows: The secondary center assigns tasks to the tertiary center according to its computing power and issues them down layer by layer; the lower-level center feeds back the computing results to the upper-level center, and finally the secondary center feeds back the computing results to the main center, which obtains the computing results through the main center.

4. A distributed high performance computing architecture according to claim 1, characterized in that: A computing capacity evaluation module is deployed in the third-level center (300) and the second-level center (200), and the computing capacity evaluation module first collects CPU usage, memory idleness and network bandwidth data in the respective centers at regular intervals; Secondly, analyze the historical data of CPU utilization to determine whether the system has insufficient CPU resources; analyze the free memory to determine memory leaks or over-allocation; analyze the network bandwidth data to determine the bandwidth usage status; Then, based on the analysis results of the performance data, the CPU usage, memory idleness and network bandwidth are optimized, and the computing power is reported to the superior center.

5. A distributed high performance computing architecture according to claim 1, characterized in that: The main center (100) and the secondary center (200) are also equipped with a task recovery module, which is used to execute: when a problem occurs in the subordinate center, the main center (100) and the secondary center (200) will recover and reset the task, thereby ensuring the normal operation of the entire system.

6. A computing method of a distributed high performance computing architecture as claimed in any one of claims 1 to 5, characterized in that: The specific steps are as follows: S1. The third-level center (300) reports its computing capacity to the second-level center (200), and the second-level center (200) reports its overall computing capacity to the main center (100); S2. The main center (100) allocates tasks to each secondary center (200) according to the computing capacity reported by each secondary center (200). The secondary center (200) allocates tasks to the tertiary center (300) according to its computing capacity, and distributes them layer by layer. S3, the third-level center (300) feeds back the calculation result to the second-level center (200), and finally the second-level center (200) feeds back the calculation result to the main center (100), and the calculation result is obtained through the main center (100); S4. During the sending process, when a secondary center (200) or a tertiary center (300) is offline, the sent task data is directly copied to the center, so that the center can assign tasks to each device node.

Images