|
|
It is important that you know the workload during normal operation in your environment. Updating an application while keeping it available means that run it on parts of your environment for a certain time.
For example, when it comes to an application rollout update, the appservers are stopped, updated, restarted on one node after the other. So for the duration of the update the remaining active servers must be able to handle the extra workload. This is a normal prerequisite for a highly available environment anyway but it might be more important during the application update because an entire node might not be available, while in a failure situation only one appserver (one JVM) might fail.
The impact is most noticeable when using a two nodes topology because the computing resources might be cut in half during the update. So if each of your systems has equal processing powers and is using more than 50% of its capacity during normal operation, then the application update might lead to overloading the remaining system with side effects such as increased response times and maybe even request timeouts for the users.
