|
|
A high availability (HA) WebSphere system is made up of two or more machines or LPARs that host appservers. Each machine or LPAR is considered a node in the WebSphere environment. Each node hosts one or more appservers (cluster members) which are interconnected in a cluster that hosts the same application. Failover and workload distribution is provided among the members of the cluster.
IBM WAS ND V6 provides a high availability framework that eliminates single points of failure for most WebSphere processes. The HAManager together with the WebSphere workload management (WLM) component provide redirection of requests within the cluster when a failure of a node or process is detected.
Refer to Chapter 2, WAS failover and recovery and IBM WebSphere V6 Scalability and Performance Handbook, SG24-6392 for more information about failover and WebSphere WLM. Refer to Chapter 6, WebSphere HAManager for information about the HAManager.
As an HA administrator, you should be aware of the different WebSphere components that participate in the HA environment and the impact when they are not available due to a hardware or software upgrade. Planning ahead for the right hardware capacity to provide high availability support and a carefully derived schedule for software upgrades are important to avoiding the risk of a failure without failover coverage.
It is imperative that the remaining nodes are capable of handling all the workload when one of them is unavailable. This way, you avoid overloading the running servers and minimize unavailability. So, a highly available environment always needs some overcapacity.
|
Perform a software upgrade when there is minimal traffic to the system and automate the process in a production environment. |
