This article describes the load balancing of the web server and the application server.
Under the idea of load balancing, multiple servers are symmetrical, each server has the same status, and can provide services separately without other servers. By load sharing techniques, the external transmission request is assigned to a certain server in a symmetrical structure in a certain rule, and the requested server is independently responding to the client's request.
A set of servers that provide services form an Application Server Cluster and provide a unified address. When a service request is sent to the cluster, select a server according to a certain rule, and the service is assigned to the server, so that the load is balanced.
By applying load balancing technology, the application service exceeds a server to provide services to limited users, and multiple servers can be used to serve a large number of users. When a server fails, the load balancing server automatically detects and stops distributing service requests to the server, and the service continues to serve the service, thereby ensuring the reliability of the service.
The above cluster techniques are generally used in web servers, application servers, etc., rather than for database servers, ie not for services with shared storage. The database server will involve a series of problems such as lockout, rollback, and more complicated. General database servers are just two machines, one of which works, and another backup. Database's dual-machine parallelism is only used in large databases. See:
System high availability and dual-machine backup common issues and scheme selection
http://www.itmgt.com.cn/ha/hafaq.htm
There are several ways to load balancing implementation:
1. The simplest is to pass DNS, but only simple turntable allocation is possible, and it cannot handle the fault.
2. If the MS IIS is based on MS IIS, Windows 2003 Server itself has load balancing services without purchasing. But this service is just a turntable.
3. Hardware mode, can be implemented by the functionality or dedicated load balancing device of the switch. There are a variety of ways for traffic distribution, but it is basically an application-independent, and the implementation load relationship with the server is not large. In addition, the price of the equipment is more expensive (the advantage is to support many servers). This method is often suitable for large flow, simple application.
4. Software mode, through a load balancing server, install the software above. This approach is flexible and the cost is relatively low. Another great advantage is that some strategies can be taken according to the application of the application and the server.
About the premium function of the load balancing is Failover, that is, a process in which the process on this server is also passed by other servers. The corresponding cost is also high, usually supported by weblogic, WebSphere software.
