Siebel Deployment Planning Guide > Siebel Infrastructure Planning >

Defining a Test and Transition Plan for the Siebel Deployment

This task is a step in Process of Infrastructure Planning.

It is important that you define a test plan to verify that the proposed deployment infrastructure functions correctly and is sized correctly. Equally important is defining a plan that transitions the Siebel deployment to production.

Observe the following best-practices guidelines for testing the Siebel deployment and transitioning it to production:

• Separate production environment. Keep the development and test environments physically separate from the production environment. Do not conduct development and test activities on the production Siebel database or, if possible, on the production database server.
• Server stress testing. Test Siebel Enterprise Server performance under average and peak workloads.

  Oracle's Application Expert Services finds that performance problems at customer sites are frequently caused by the following:

  • Servers were tested at much less than average or peak workloads. This prevents configuration and tuning problems from being uncovered.
  • Siebel Server components are either incorrectly distributed across servers or are not configured correctly.
  • The load balancing strategy is ineffective under typical workloads. This can be caused by stress-testing the servers with a workload that has characteristics different than the production environment.
• Failover and resiliency testing. Define a test plan that evaluates the effect of server component failures. Work closely with your implementation team to identify all of the components that could represent single points of failure, as noted in Defining High Availability Policies.

  Define a server cluster test plan that evaluates failover behaviors. Run the test plan under average and peak workloads. It is particularly important to verify that failover performance under peak workloads is acceptable.

• Database server testing. Define a test plan that evaluates the following:
  • OLTP performance. Consider OLTP performance under average and peak workloads.
  • Database server platform failover. Typically the database server is clustered.
  • Recovery from database corruption. The database vendor usually provides recovery mechanisms.
  • Batch processing support. Verify that the database server correctly handles batch jobs from servers as well as synchronization requests from Siebel Remote components.
  • Web Client users. Verify that batch jobs do not degrade transaction processing performance and are completed in a reasonable time.