Disaster recovery testing should be designed to test that your solution can cope with a disaster scenario. Disaster scenarios range from minor hardware failures to the loss of your production site. When designing your solution, you must consider the myriad possible failures and have a design or strategy to cope with those.
The disaster recovery test phase is when you get to validate that your solution works as designed under a failure scenario. The following are some of the scenarios that you should consider testing. They might not all be appropriate for your solution; nevertheless, they should trigger some ideas about what you need to test. When thinking about all these types of tests, it is important to remember that failures generally happen under load; therefore, these tests should be carried out when the system is under load, and you should ensure that no data is lost:
Database cluster failover — For most BizTalk deployments, the database is deployed in an active-passive cluster. A cluster may failover for a number of reasons, and when it does the chances are you will be processing messages. You may have custom code that is accessing a custom database. You should also ensure that it behaves as designed in a cluster failover scenario.
Processing node failure — All BizTalk deployments should have at least two instances of each host instance running on different physical servers. BizTalk is designed to naturally load balance across multiple instances ...