A storage architecture system based on all-shared switching

Abstract

The invention discloses a storage architecture system based on full sharing and exchanging. The system comprises at least one main engine adaptation module, at least two main engine exchanging devices, at least two controllers, at least two controller exchanging devices, at least two storage exchanging devices and at least one magnetic disc, wherein each main engine adaptation module is connected with all main engine exchanging devices; the all main engine adaptation modules and the all main engine exchanging devices form a main engine layer; each controller is connected with all controller exchanging devices; the all controllers and the all controller exchanging devices form a controller layer; each storage exchanging device is connected with all controllers; and all storage exchanging devices and all magnetic discs form a storage layer. The system has the advantages of reducing time delay of the system, enhancing reading and writing performance of the system and improving reliability of the system.

Description

technical field [0001] The invention relates to the technical field of storage architecture, in particular to a storage architecture system based on all-shared switching. Background technique [0002] At present, the storage architecture has achieved rapid development in the past few decades. Storage was first accompanied by the evolution from mainframes to minicomputers, and has a history of 40 to 50 years. However, the high-end storage system of the open system has only been around for 20 years. In chronological order, from the bus architecture to the switched architecture, the matrix direct connection architecture, and the distributed architecture. The bus switching high-end architecture is not very scalable, and because it is based on the bus, there is bus contention. The switched architecture connects the front-end and the back-end for data exchange through the switching ASIC. Compared with the matrix direct connection, it has better scalability, but the problem is tha...

AI Technical Summary

Problems solved by technology

However, due to the asymmetric distribution of distributed architecture resources, there are application limitations in actual use, see figure 1 , the controllers in the controller layer in the distributed architecture are grouped in pairs, and then connected to the corresponding disks in a combined form to complete the data channel connection. Blocks 1 to 6 on the controller in the figure represent Different ports of the controller, but this kind of connection will cause some data to fail to reach the disk box it needs to reach. For example, the place where the written data should be written is the disk box corresponding to the group of controller 7 and controller 8. , but the data to be written is transmitted to controller 1 or controller 2, then the written data cannot reach the corresponding disk box of controller 7 and controller 8

[0004] This distributed architecture will cause some data to fail to reach the disk box it needs to reach, and the delay time of data transmission will increase, that is, the system delay will increase, and the data will not reach the corresponding disk box, and cannot be stored in the corresponding disk box. If the data is written on the computer, the distribution of system resources is asymmetrical, and there are application limitations in actual use, resulting in low system output performance and reduced reliability.

Images