The load-balanced topology consists of a highly available
application server cluster, an HTTP server cluster, a database cluster,
and a directory server cluster.
In a load-balanced topology, services are highly available
because all components are either in active-active mode or in hot-standby
mode. A high volume of users can access the system simultaneously,
with minimal impact on operations. This topology is commonly used
because it also addresses performance and scalability of the infrastructure,
and it can be used with the advanced configurations.
The system
environment consists of the following highly available components,
which can be complemented by cluster management software to automate
failure detection and service startup:
- Failover of WebSphere® Application Server is
managed by WebSphere Application
Server Network Deployment.
Configuration data is replicated to every node agent in the cell,
so the impact of a deployment manager failure on the application client
processing is limited. When the deployment manager server becomes
available again, all node agents in the cell reactivate and service
returns to normal.
- The HTTP server is supported by a standby server, which becomes
active when the primary HTTP server fails.
- The database server can be automated to run on a primary database
server and a standby database server. Automated failover and data
integrity is handled by DB2® HADR.
If the primary database server fails, the services are automatically
restored on the standby database server.
- Tivoli® Directory Server can
be configured in a load-balanced environment by introducing a highly
available proxy server. In this scenario, both Tivoli Directory Server instances
are running and synchronized through peer-to-peer replication. If
one of the directory servers fails, the proxy automatically directs
the requests to the next active instance. A primary server and a standby
proxy server are configured in a warm standby configuration with highly
available shared storage. The shared disk provides a common installation
directory and is mounted only on the active server. If the primary
proxy server fails, the standby server can be automatically started
up by the cluster manager. If one of the directory servers fails,
the proxy automatically directs the requests to the next active instance.
The following diagram shows a load-balanced topology,
where a product such as Maximo® Asset Management
is deployed in a scaled WebSphere Application
Server Network Deployment environment.
Standby
servers provide failover support for each component and a highly available
shared file system is used to store attachments and other files. As
individual components, the HTTP server, the database server, and the
directory server are all highly available. WebSphere Application
Server Network Deployment manages
the application server nodes in the system.
