A method to realize cluster server inner load sharing agent

Abstract

The invention discloses a method for realizing the loading sharing agent in a cluster server, comprising the following steps: a client sends a node server selecting request to a agent server; the agent server returns a list of selectable node servers to the client according to the loading situation of the cluster server; according to the list of selectable node servers, the client determines the selected node server and establishes connection directly with the selected node server by negotiating the loading sharing algorithm with the agent server until the conversation between the client and the node server is ended. As adopting the loading sharing agent method that the client is provided with the optimal server node and the suboptimal server node, the invention strengthens the external stability of the cluster server. Furthermore, the agent server does not retransmit the application data between the client and the server; therefore, the agent server is not easy to cause the network bottleneck.

Description

technical field [0001] The invention relates to the fields of computer application and communication, and more specifically, to a method for realizing load sharing agents in cluster servers. Background technique [0002] The rapid growth of the Internet has led to a rapid increase in the number of visitors faced by various network servers and communication servers, and the server has become overwhelmed. If a large number of user service requests cannot be processed in time, the server will be interrupted. In the past, the way to solve these problems is to replace the original server with a server with stronger computing power. Even so, the load capacity of a single server is limited, and it is impossible to expand infinitely. The performance has shown an exponential increase. Therefore, the system model of load sharing composed of multiple cheap servers has gradually become the mainstream, thus expanding the load sharing technology. The load sharing system mainly distribute...

AI Technical Summary

Problems solved by technology

In the early days of load sharing, many networks were designed using domain name rotation, that is, one domain name corresponds to multiple servers. As a cheap solution, domain name rotation can solve the load problem of some servers. However, Due to this load sharing method has great limitations: it is impossible to send user service requests to different servers according to the load conditions of each server; when one of the servers fails to work, the system will still send the User access requests are sent to the faulty server, causing interruption of some services; because domain name interpretation generally has caches on servers in various places, user access requests in a region will be concentrated on the same server

This traffic load sharing method dynamically selects links according to the busyness of the channel, which can be applied to traffic load sharing at different levels and applications of data communication equipment, but this traffic load sharing also has shortcomings, such as: client to server The request data cannot be the same every time, and the server found by the proxy server is different every time, so the consistency of the session between the client and the server cannot be guaranteed at all. Moreover, since the client requests the server every time, the proxy server must Performing flow calculations for it wastes a lot of time and a lot of proxy server resources, resulting in a heavy load on the proxy server

Images