We use Link Tuner from Fishbowl to automatically run a test with a certain
sized model (small, medium, or large) on a certain number of PCs (12 or
27) to find, add to workspace, open in CREO, then close CREO as many times
as possible in 2 hours to load test our system. We then count the number
of times the script completes. The more times it completes, the faster
the throughput. We can also run diagnostics on the network during this
testing, as well as use the PSM to get more information. In your case,
something like this would be good to run it locally, then run it remotely,
see how much different it is, then start peeling back all the different
layers of the problem and optimizing where possible to get closer to what
you want in terms of performance. You might try using remote blade
servers to run CREO locally while remote sites only see visual
information. You might try Riverbed or something similar. You might try
making a replication server for remote locations.
The point is that this lets up flood the system to capacity, and see
what's happening over an extended period of time.
If the problem is purely "check in is slow everywhere" rather than just
"check in is slow from a remote location" then you should use a different
approach by using the PSM with PTC tech support to optimize the checking
code, probably using DB indexes or something.
Al
