Highly Available WebSphere Business Integration Solutions

34 Highly Available WebSphere Business Integration Solutions

3.6 HA cluster design

By using HA clustering software along with the WebSphere Business Integration

portfolio it is possible to enhance the availability of the services that the portfolio

is providing. With a suitably configured HA cluster, it is possible for failures of

power supplies, nodes, disks, disk controllers, networks, network adapters,

queue manager processes, or other component processes to be detected and

automatically trigger recovery procedures to bring the effected WebSphere

Business Integration component back online as quickly as possible.

Although each vendor adapts their clustering technology according to their

particular hardware architecture, there are many common procedures that are

shared among the different clustering solutions. This is also true in the case of

implementing MQ, Message Broker, and InterChange Server within a given

clustering solution. For instance, no matter what clustering approach is being

used, MQ, Message Broker, and InterChange Server should have their own

product files located on local disks and data and log files resident on a shared

storage device. The same rule applies to database management systems that

are part of the WebSphere Business Integration environment such as the

Message Broker database or InterChange Server’s repository database. The

setup described is required in order to facilitate takeover of resources by the

acquiring node; this will allow for the WebSphere Business Integration services

to be restored.

All clustering approaches provide different mechanisms to define resources that

are going to be managed by the HA clustering software. Examples of these

resources are: A particular queue manager, a database instance, all broker

processes (including execution groups), or all user name server processes.

Within the HA clustering environment, related resources are grouped together

into resource groups

2

.

Resource groups

Resource groups are the unit of failover in HACMP For example, a resource

group is migrated as a unit when failover occurs. A resource group is a unit that

contains dependent resources: IP labels, applications, file systems, and volume

groups. A resource group has attributes that define how it behaves during node

startup, resource group failover to another node in the case of application or

node failure, and when the resource group will fallback to the reintegrating node.

Ideally, a resource group will contain only the resources necessary to define an

operational unit. This approach optimizes flexibility in failover, and minimizes

disruption during a failure or planned maintenance.

2

The term resource group is HACMP specific. For other equivalent terms, please see “Resource

groups” on page 16.