Research And Implementation Of High Available Load Balancer In Cluster System

Cluster system shares resource across network. It not only uses existing computational resource well, but also can reach high performance with lower cost of software and hardware. As a significant content of Cluster research, High Availability (HA) aims to prevent whole system from crush caused by single point failure by means of redundant software and hardware resource. Providing Non-Stop service is the goal of HA.This paper analyzes the basic technique of Cluster and gives a HA expressions for Cluster system by Stochastic Process Theory. We emphasize on the HA technique of Checkpoint and Process Migration. In order to validate the feasibility that Checkpoint and Process Migration can be implemented on general OS using Kernel Module, we design and implement CKPM. This is a significant test on HA by software.Based on Checkpoint and Process Migration, we give a Cluster Balancer HA resolution. Then we implement a prototype based on LVS.The paper makes a series of researches on Checkpoint and Process Migration and can help the design of HA Cluster system.