Highly available message transmission frame and method oriented to distributed file system

Abstract

The invention provides a high usable message transmission frame aiming at distributed file systems and a method thereof. The message transmission frame comprises a file system interface layer which provides an access interface for a file system layer, a file operation processing layer which is respectively alternated with the file system interface layer and a bottom layer protocol packaging layer, and is used to process requests submitted by the file system layer and to submit the operations to the bottom layer protocol packaging layer, and the bottom layer protocol packaging layer which is communicated with the network protocol layer and is used to pack the operations submitted by the file operation processing layer and then to send a message through the network protocol layer. The high usable message transmission frame and the method thereof can provide network usability guarantee for the distributed file systems equipped with one set of networks or multiple sets of networks.

Description

technical field [0001] The present invention generally relates to a message transmission framework and a message transmission method of a distributed file system, and particularly relates to implementing a high-availability mechanism at the message transmission level of a distributed file system to improve the availability of file data network transmission, thereby improving the overall availability of the distributed file system Message transmission framework and method. Background technique [0002] The cluster system has become the mainstream architecture of super servers because of its good scalability, high cost performance, high availability, inheritance of existing hardware and software resources, short development cycle, and good programmability. Since the cluster system was born in the 1970s, it has experienced the development from a dedicated cluster system to a general cluster system, and its scale has grown from dozens of processors to tens of thousands of proces...

AI Technical Summary

Problems solved by technology

[0005] Since the file system is located on the underlying network protocol stack, and the distributed file system operation is not transactional (composed of multiple network transmission operations), the fault tolerance mechanism of the underlying network protocol layer cannot provide the required availability guarantee for the file system

However, network faults are directly solved at the file system layer, and the file system operation and network fault-tolerant mechanism are closely combined to solve software network faults, but it is not easy to recover from network hardware faults, and this approach makes the file system and the underlying network protocol more coupled. It is not conducive to expansion and transplantation; network hardware topology fault tolerance is based on a network topology model with high redundancy, which to a large extent solves network failures caused by hardware, and is especially effective in dealing with network segmentation. The problem is the high hardware cost and difficulty in engineering implementation, and it cannot handle well the network faults when the hardware system is running well, such as sudden drop in network performance, logical connection interruption and other abnormalities

[0006] In addition, in terms of solving network fault tolerance at the message transmission layer of the file system, it implements request retransmission through a state transition machine. Although it can solve software errors such as connection interruption, it cannot solve hardware failures such as network segmentation.

Images