The next step here is to differentiate between application problems and enterprise server infrastructure problems.
Application failures other than abends are often application problems. Look at the active trace dataset coupled with the system trace table in the dump X dataset (casdumpx.rec) to establish the application activity before the failure.
Application abends can be due to application problems or enterprise server infrastructure problems. Look in the active dump dataset for an RTS error dump. If found, evaluate the RTS error and classify the problem from this. What module experienced the failure, application or enterprise server infrastructure?