08-13-2009 01:21 PM
Our organization uses a 100% VMWare environment. For the most part, this has worked quite well, but I have noticed that certain portions of the Enterprise Vault roles do not function well in a virtual environment. The current configuration is below. All servers are Windows 2003 SP3 32-bit. The roles and general virtual configuration is listed. Each EV# is it's own server. The environment is approximately 5000 users. We have gone through the performance guide several times and have made a good amount of the recommendations as applicable.
Servers:
EV1 - Mailbox Archiving
4 procs (these 4 cores have been dedicated in VMWare for this virtual machine), 4 GB RAM, 5 index locations each on it's own LUN on NetApp, Vault Cache on dedicated drive on NetApp
EV2 - Journal Archiving
4 procs, 4 GB RAM, 1 index location on NetApp
EV3 - Directory and PST Location (Locating only, no collection or importing)
2 procs, 4 GB RAM
EV4 - Discovery Accelerator
2 procs, 4 GB ram
EV5 - SQL 2005 32-bit
4 procs, 4 GBs, dedicated drives for Logs and Data (seperated)
Current Issues:
- EV1 is completely running non-stop 100% all four procs 24 hours a day. Indexing issues that require restarting of services approximately 2-3 times per week.
- EV4 and EV5 - The enabling of analytics throws all four processors on EV5 to 100%. The creation of full text XML files from this causes extreme performance hits. Estimated time to complete 100,000 items for analytics is 8.4 hours (possibly longer).
Questions:
- Does anyone have a similar size environment and uses VMWare?
- Has anyone been able to enable analytics successfully within a VMware environment? How large was the case? Did you have the same problem with the full text searching?
- I would love to get a definitive answer on VMWare environment and Enterprise Vault. I only can find general recommendations but not a specific, "hey, with this number of mailboxes you will have this much lost in a VMWare environment" statement.
Any assistance would be greatly appreciated.
Thanks.
Solved! Go to Solution.
08-27-2009 04:45 PM
08-13-2009 09:36 PM
08-14-2009 04:32 AM
08-14-2009 09:28 AM
For EV1, the top two processes are StorageCrawler and RetrievalTask. The Systems is the third for the most part that keeps fluctuating in the list. MigratorService and IndexServer is in the top 5 to 10 mostly but it fluctuating.
The schedule for archiving is basically from 4 PM to 4 AM for mailboxes.
As for the SQL server, the server does host both the DA databases and the standard EV databases. I was incorrect in the memory count, it's actually been giving 16 GBs of RAM. Again, this is a virtual machine.
Are there particular disk transfers that would be best to look at to get a better idea on the issue?
Thanks for all of your help thus far.
08-14-2009 10:13 AM
08-27-2009 03:58 PM
08-27-2009 04:45 PM
08-28-2009 11:59 AM
Yeah, we are on 8.0 SP2 and also have implemented a hotfix for Discovery Accelerator because of a scheduled search issue that exists.
I will check those out and see if I can find something else that I am missing.