Multi-Instance Upgrades

The sequence of upgrade procedures in an upgrade plan depends on how redundancy is being used in a deployment architecture. Multiple instances of a Java ES product can be used to achieve high availability, scalability, serviceability, or some combination of these service qualities. Three technologies make use of redundant products in Java ES deployment architectures: load balancing (Directory Proxy Server, Web Server, Web Proxy Server, Application Server, Access Manager, and Portal Server), high availability techniques (Sun Cluster and High Availability Session Store, and others), and Directory Server replication.

In most cases where redundancy is involved, upgrades must be performed without incurring significant downtime. These rolling upgrades attempt to successively upgrade redundant instances of a product without compromising the service that they provide.

Redundant instances are usually deployed across multiple computers. For upgrade planning, you might need to isolate the upgrade of replicated products from other product upgrades in order to achieve minimal downtime. In such cases, you often perform all the pre-upgrade tasks for the replicated products on each computer before performing the rolling upgrade.

Each replication technology has configuration or reconfiguration procedures that might affect the overall sequence of Java ES product upgrades. For example, products that run in a Sun Cluster environment can require upgrading Sun Cluster before upgrading the products that are running in the Sun Cluster environment.