The Design And Implementation Of Load Balancing System For Enterprise Data Center

Every time the Internet technology innovation puts forward higher requirements to the network infrastructure.When a large number of user access to the server,the server's rapid response and low latency network can provide users with a good experience.A single server obviously can not meet the company and the organization's business growth needs.If the server needs to maintain or upgrade,the services will be suspended.The emergence of server cluster technology can solve this kind of problem,and load balancing technology is an important part of server cluster technology.A load balancing system based on epoll model is designed to meet the increasing demand of data access in enterprise data center,which traditional load balancing system can not meet the requirement of high concurrent access and stability.In order to manage system resources flexibly,polling technology has been used to repackage the system memory and thread resources.The memory pools and thread pools make resource be allocated and released more secure and convenient.Referring to the operating system knowledge on the asynchronous I/O design,user requests are handled by a similar process.In addition,in order to alleviate the pressure on the server,the system uses content caching technology to cache the response content.At the same time,the traditional Weighted Round-Robin algorithm is optimized,using a smooth Weighted Round-Robin algorithm,so that the server can be hit smoothly.Web technology has been applied to provide an interactive interface to distribute system configuration.And the statistics of virtual server traffic can also be displayed.The system is deployed in an enterprise data center and acts as a "traffic policeman" for the cluster.It can distribute high concurrent requests to the back-end server nodes,making load of the cluster more balanced.In addition to providing stable request distribution and feature support,the traffic can be intelligently managed.