Community Tip - Your Friends List is a way to easily have access to the community members that you interact with the most! X
Hello Gurus,
We are planning to move our Windchill Servers from Physical servers to Virtualization servers. Any inputs about the issues/challengesthat have been faced will be of great help.
As far as my experience with Windchill and VM goes the main issue for any Global Company is the Network Latency.
Any ideas and experiences that you have and can share will be of great help .
We are currently on Windchill 9.1_MO50.
Thanks
Mayur
Hello,
I know from other customers that Windchill can be made to work successfully on Virtual servers, but our experience was not good. We used Virtual servers for our Production hardware “application tier” from 2005 till earlier this year, we still use them for our Prototype and Development environments.
We run a clustered environment (have since 2002, on a variety of operating systems through the years) with a number of user facing slave “application servers” (running servermanger, methodserver, webserver and servlet) behind a load balancing router, with a non-user facing physical server acting as the cluster master, all connecting to a physical database machine with external storage. We have 2000 active users.
For a time we had physical and virtual machines in the Production application tier so we could make direct comparisons between them, we could never get the virtual machines to perform nearly as well as the physicals under heavy load. For light user load or a single heavy task like when loading a large file they could get close to the physical machines performance. But load from typical user activity consumed significantly more CPU resources on the virtual machines, despite identical spec and Windchill configuration. We could have lived with the discrepancy in performance, but it appeared to cause reliability issues with the Windchill application stack. During unexplainable peaks of sustained CPU utilisation on the Virtual servers either methodservers would randomly drop out, or Apache.
We did invest a lot of time investigating the issues, and working with our hardware team. Following all the recommendations from PTC and working with tech support. Finally this year we replaced the Production virtual servers with physical machines. For everyone involved that was a great relief, to quote one of our users, “Windchill is working like never before”.
From an Application owner perspective, virtual machines are as much work to maintain as physicals, the application needs installed, configured and maintained. They can never perform as well as an equivalent physical machine, are unpredictable under heavy load, and introduce an extra level of complexity to your Windchill installation. Windchill is already complicated enough, particularly when troubleshooting performance issues. On the flip side it is easy to clone servers, experiment with changes and roll them back. You also get some fault tolerance on the hardware, but the recommendation is to have a physical database machine so I am not so sure that is really much benefit.
If you do go virtual my advice is to work with your hardware team, carefully follow all the recommendations from PTC and try to do some kind of load test before you go to production. Hopefully someone who has had a positive experience will post and provide some balance for you.
-----
Lewis