All,
We had a situation where, after some DHCP service work, the routing from the users PC to the Windchill server used our MAN (Metropolitan Area Network), but the return route from the Windchill server back to the users PC was using our WAN. 5 hops from PC to WC, 10 hops from WC back to the PC.
This caused massive instability in working in Creo Parametric linked to Windchill. Users were crashing, hanging, etc. over 10 times a day doing a variety of tasks in Creo whether related to WC functions or Creo functionality. i.e. editing a sketch vs. doing a save & upload.
Our normal LAN only, MAN only, and WAN only users are not having any issues. It's just those folks using the MAN out and WAN back.
As expected, the WAN back route does take longer, but it's not excessive (traceroute hops ~5ms back vs. < 1ms for out). However, traceroutes are simply a snap shot in time, and not really reflective of the overall network performance. The WAN back routing could be excessive at the time the users crash. That still doesn't explain why our normal WAN users aren't crashing as well.
We fixed the issue by forcing the local subnets to use the MAN only.
Does WC expect the route time out and back to be close? i.e. 10 ms out and 10ms +/- some percentage back? If so, what would control this in WC?
As a more general question, what WC settings would help with "unstable" networks where the route times may vary randomly throughout the day? Is there some threshold of network performance (latency, # hops, etc.) that would dictate changing the WC and/or client setup? We'd rather live with somewhat slower but stable performance vs. crashing and losing work and schedule time.
We've setup WC and the clients per the PTC documentation for working with CAD data already; No local cache compression, IE settings, etc.
Pro/Intralink 10.1 M030, Creo Parametric 2.0 M060 and M110.
Regards,
Daniel Nordin
BAE Systems, Information Management
Business Technology and Transformation
Engineering Tools, ME Support Group
Nashua, NH 03061
-
