Zookeeper-based request coordination method and device, computer equipment and storage medium

Abstract

The invention discloses a Zookeeper-based request coordination method and device, computer equipment and a storage medium. Among them, the Zookeeper-based request coordination method includes: receiving the write request from the client; mapping all service nodes in the Zookeeper cluster to the node ring; calculating the distance between every two adjacent nodes on the node ring through the proxy node, and obtaining the leader according to the distance Node; query the routing table through the proxy node, and submit the write request to the leader node for processing. According to the embodiment of the present invention, the requirement for the number of Servers in the leader election of the existing ZooKeeper cluster is made up, and the horizontal expansion capability and fault tolerance of the ZooKeeper cluster are improved; the agent node query routing table is introduced to quickly route the client request, avoiding the existing client Inefficiency problems caused by writing to the main mode or data consistency problems caused by writing to arbitrary modes.

Description

Technical field [0001] The present invention belongs to the field of distributed system cluster, and more particularly to a request coordination method based on ZooKeeper, a request coordination device, computer device, and computer storage medium based on ZooKeeper. Background technique [0002] ZooKeeper is a software that provides consistency services for distributed applications, distributed applications can implement configuration management, name service, distributed lock, cluster management, queue management, and more. [0003] ZooKeeper technology in the related art has the following defects: one aspect, ZooKeeper's leader node (Leader) election is based on a messaging consistency algorithm (PaxOS algorithm), and the PaxOS algorithm acquires most Server support in a service node (Server). In the case, it can be recommended as Leader, so the process is a repeated process, making the number of Server participating in the elections not too much, and the total number of Serve...

AI Technical Summary

Problems solved by technology

[0003] The ZooKeeper technology in the related art has the following defects: on the one hand, the election of the leader node (Leader) of ZooKeeper is based on the message delivery consensus algorithm (paxos algorithm), and the paxos algorithm obtains the majority of Server supported by a certain service node (Server). Therefore, the process is a repeated recursive process, so that the number of Servers participating in the election cannot be too many. At the same time, the total number of Servers must be an odd number 2n+1, and the number of surviving Servers must not be less than n+1, which limits the scalability and fault tolerance of ZooKeeper

Or only select m servers from n servers to run the ZooKeeper service to ensure that the data of the ZooKeeper service will not be migrated when the entire cluster is restarted or more than half of the servers where the ZooKeeper service is located are down, but in this case, there will be n-m servers. In the standby state, the processing capacity of the entire server cluster is only all m / n, resulting in waste of resources

Or choose the node with the smallest number as the Leader node every time, but this situation does not consider the load of the node and achieve load balancing

[0004] On the other hand, the read and write access of the client has two modes: write-master and write-arbitrary. In the write-master mode, the modification of the data by the client is submitted to the Leader, and all write requests submitted to the follower (Follower) client It needs to be transferred from Follower to Leader for processing, and the processing efficiency is not high; when writing any mode, the modification of data by the client can be submitted to any Server node, and there is a risk of data consistency and deadlock

Images