High availability for HTTP servers

An HTTP server that operates as an individual component can lead to a single point of failure. You must configure the HTTP server to integrate it into the high availability topology and to minimize the risk of a system failure.

A highly available HTTP server is installed on two nodes in a cluster; a primary node and a standby node. If the primary node fails, the HTTP server operations can be restored on the standby node immediately.

You can use System Automation for Multiplatforms to automate the detection of HTTP server failure and to trigger failover to the standby node. HTTP server installation data can be stored individually on each server or by auto-mounting a shared installation directory on the active node only.

Some organizations do not use an HTTP service IP, but instead use an external load balancer to process HTTP requests. In such a scenario, you can have multiple active HTTP servers with the external load balancer that directs the requests. If you use an external load balancer, ensure that the load balancer does not represent a single point of failure.

The following diagram shows a highly available HTTP server cluster, in which a primary standby HTTP server is supported by a standby HTTP server:

The diagram is described in the main body of text.

For more information about highly available HTTP servers, see the HTTP session failover section of the WebSphere® Application Server Network Deployment V6: High Availability Solutions IBM® Redbooks® publication.



Feedback