Method and device for writing data in distributed system

Abstract

The invention discloses a method and device for writing data into a distributed system. When data is requested to be written into the distributed system; the main node can detect whether the performance of each standby node meets the requirement or not; for a standby nodeS with relatively low performance, the main node can temporarily ignore the feedback of the standby nodes; only the write-in success feedback of the standby node of which the performance meets the requirement is waited, namely, as long as the standby node with the performance meeting the requirement returns a response message,the main node can feed back writing success to the terminal; compared with whether a standby node with poor performance exists or not, the main node feeds back a successfully written distributed system data writing mode to the terminal only after all the standby nodes feed back a successfully written response message. The problems of long time and low efficiency of feedback writing success of thewhole distributed system due to slow feedback of the standby node with low performance can be effectively avoided, so that the data writing efficiency of the distributed system is improved.

Description

technical field [0001] The present application relates to the technical field of communications, and in particular to a method and device for writing data in a distributed system. Background technique [0002] A distributed system is a high-performance system that can provide multiple copies of fault tolerance. Typically, strong consistency is maintained across multiple copies of data in a distributed system. When receiving a write data request, the distributed system will report that the write is successful only if all replica nodes (that is, all standby nodes) are successfully written. However, due to the high usage of the CPU (Center Processing Unit) by operations such as flashing disks, network jitter, and other reasons, some nodes may experience short-term performance degradation and cannot write data in time. The latency of data writing on these nodes will be extended. And because the data writing time on these nodes is too long, the delay time for the distributed s...

AI Technical Summary

Problems solved by technology

[0003] The embodiment of the present application provides a method and device for writing data in a distributed system to solve the problem of too long data writing delay time in the distributed system caused by the performance jitter of individual nodes, thereby improving the data quality of the distributed system. write efficiency

In this way, through the performance detection to determine the standby nodes that need to ignore the feedback and need to wait for the feedback, the master node can effectively avoid the problem of long and inefficient feedback writing success in the entire distributed system due to the slow feedback of the standby node with low performance. Thereby improving the data writing efficiency of the distributed system

Images