WebSphere WebSphere Application Server Network Deployment, Version 6.1.x Operating Systems: AIX, HP-UX, i5/OS, Linux, Solaris, Windows, z/OS

Service integration high availability and workload sharing configurations

The configuration of service integration is very flexible. There can be individual messaging engines, or clusters that contain multiple messaging engines, where those messaging engines can share workload, be highly available, or do both.

The type of bus member you create affects the options you have available to you. You create a bus member by adding either a server or a cluster to a bus. If you create a server bus member, a single messaging engine is created automatically. For many purposes this is sufficient, but such a messaging engine is a single point of failure and cannot share workload, so has no high availability or workload sharing characteristics.

If you want to avoid a single point of failure by allowing failover, or if destinations need to be partitioned to allow sharing of the messaging workload, use a cluster bus member. With a cluster bus member you can achieve high availability, workload sharing, or both, depending on the policy you configure for the messaging engines in that cluster. See Bus member types and their effect on high availability and workload sharing configuration for more information about server and cluster bus members.

After you create a bus member, you configure a policy to control the availability behavior of the messaging engine on that bus member. If you want high availability, you can use a cluster bus member with one messaging engine and the default service integration policy, "Default SIBus Policy", which allows the messaging engine to fail over to any other application server in the cluster. Alternatively, you can create a new policy and configure it to specify other availability behavior, such as a preference for particular servers or the ability to fail back.

If you want workload sharing but not high availability, you can use a cluster bus member with multiple messaging engines and create a Static policy for each messaging engine. This could be useful for scalable express messaging, in which there is no persistent state associated with any one messaging engine, so no failover is required.

If you want the messaging engines to be managed by an external high availability framework, create a "No operation" policy for them.

For more information about policies and configuration, see Policies for service integration.

With the options just described, it is possible to achieve high availability, workload sharing, or both. The following table shows how you can achieve different configurations. See the related links for some examples of what you can configure.
Table 1. Service integration configurations
Configuration Type of bus member Number of messaging engines Policy type
Simple Server 1 Default ("One of N")
Simple Cluster 1 Static
High availability Cluster 1 "One of N" or "No operation"
Workload sharing without high availability Cluster more than 1 (typically, one messaging engine for each server) Static
High availability and workload sharing Cluster more than 1 (typically, one messaging engine for each server) "One of N" or "No operation"
Related concepts
WebSphere Application Server high availability
Multiple-server bus with clustering
Related tasks
Configuring high availability and workload sharing of service integration
Creating a policy for messaging engines
Configuring a policy for messaging engines
Using match criteria to associate a policy with a messaging engine
Configuring a Static policy for service integration
Configuring a One of N policy for service integration
Configuring a No operation policy for service integration
Configuring shared durable subscriptions

Concept topic

Terms of use | Feedback


Timestamp icon Last updated: 26 February 2009
http://publib.boulder.ibm.com/infocenter/wasinfo/v6r1/index.jsp?topic=/com.ibm.websphere.pmc.nd.multiplatform.doc/concepts/cjt0007_.html

Copyright IBM Corporation 2004, 2009. All Rights Reserved.
This information center is powered by Eclipse technology. (http://www.eclipse.org)