How can I configure Dynatrace so that it warns me in case not all application instances in an agent group get a simular (CPU, I/O, ...) load?
I don't think there would be a way to actually base an incident off of whether instances are reporting different metrics. There are incidents for when one of those incidents gets to an unhealthy stage so that it will let you know if instance X is unhealthy but not just that they are not all the same.
Is there some issue in particular that you are trying to watch for?
We have 6 instances of webMethods that execute behind a loadbalancer. One of the node was constantly consuming a minimum of 30% CPU, while the other dropped to lower % of CPU when there was less load. Restart of this node 'solved' the issue. So it would be good if Dynatrace could warn me when this happens again.