Even if environments are well matched, testing may miss issues for reasons such as the following:
-
Custom-developed applications are often not tested for impact, memory stability, or session and license leakage.
-
Testers almost never have time to test every single activity. Even if they do, it is difficult to know what every activity is and the right order in which to perform the activities.
-
If patches are applied, components change, or business processes change mid-testing, the thoroughness of the coverage of the test changes.
-
It is very difficult to replicate real world concurrency. Sometimes issues have occurred that were caused by simultaneous usage of different modules.
-
Issues are sometimes related to the status of transactions that are already within the system. When a test system doesn't reflect the historical data that interacts with transactions, possible issues may not be observed during testing.
Testing can only provide a sampling of potential risks and issues. Testing greatly reduces the risk of an upgrade, but it can't remove all risks associated with upgrading. For this reason, managing risk requires more than just testing. Incremental Parallel Upgrades help manage additional risks.