Soft load balancing implementation scheme based on API

Abstract

The invention provides a soft load balancing implementation scheme based on an API (Application Program Interface), which mainly comprises a communication module and a detection module which are embedded into a client as a whole. In general, the client side initiates a request and receives a response by calling a synchronous or asynchronous interface of the API, and a communication mode of TCP synchronous or asynchronous short connection is adopted between the API and the cluster server side. And the API can regularly detect the survival state of each node in the cluster server and obtain cluster IP configuration and a load balancing strategy thereof. The implementation scheme has the advantages that hardware equipment support is not needed, and the cost is effectively reduced; a synchronous communication mode and an asynchronous communication mode are supported, and a communication module is integrated in an API and is transparent to a client; load balancing is supported, and load requirements of personalized load factors are met; non-perception of the expansion or contraction client is achieved, and the elastic expansion and contraction function is met.

Description

technical field [0001] The invention relates to the technical field of load balancing, in particular to an implementation scheme of API-based soft load balancing. Background technique [0002] Load balancing is a very important link in the cluster system. Requests are dispatched to several nodes in the network for processing through load balancing to ensure high availability of the cluster system. Although hardware devices such as F5 can solve load and expansion problems well, they are expensive, and F5 itself has a single point of danger. The soft load is favored by users for its low cost and convenient function development. [0003] At present, there are various modes for implementing soft load. There are not only open source framework products, such as registration center-based ZOOKEEPER, but also various self-designed and developed load balancing products or strategies. [0004] The disadvantages of the ZOOKEEPER load balancing solution based on the registration center...

AI Technical Summary

Problems solved by technology

[0004] The disadvantages of the ZOOKEEPER load balancing solution based on the registration center are: 1) The registration center is a centralized node, which is contrary to the concept of decentralization; 2) The cluster server registers on the registration center, and the registration center will register the cluster server The information is sent to the client. If the registration center is abnormal, the client may not be able to obtain the information of the cluster server in time; 3) The load and communication functions between the client and the cluster server need to be coded by the client, and the integration capability of the client is required higher

Disadvantages are: 1) You need to install an independent module on the client, and you need to start the relevant detection process; 2) If the detection process fails, the available address list is not up-to-date, which will lead to transaction failure; 3) Cannot dynamically perceive changes in cluster server nodes , need to maintain the cluster server address list on the client

Images