jshem wrote:
Final result:
We reported the problem and got an official response I'm very unsatisfied with.
I provided a SQL trace of 7 minutes of processing that generated 180MB of transaction logs. But it appears they didn't look at it.
They said to:
* Check the SQL Server Version is supported - standard stuff
* Shrink the Transaction Log File - Useless advice, we explained in our report that the MDF and LDF file sizes were not an issue. we were doing TRN backups every 15 minutes....it was the sheer volume of transactions and subsequent large backup files that was the issue.
* Finally they said to check SIMPLE vs FULL recovery mode...We were told to use SIMPLE. Well in our report, we explained we were using FULL because we wanted point-in-time recovery. But they avoided that completely by telling us we must use SIMPLE.
Result, I didnt want to argue with VMWare since our Sysadmins were willing to give up point-in-time recovery for a Differential backup every 6 hours and the ability to recover back to then...they seemed to be ok with potentially losing 6-hours of stats, etc...
But whoever sees this, I hope you understand that VMWare did not really understand the issue: Why is their application generating so much TLOG activity?....and that they are in effect stating they do not support SQL Server's ability to restore to a point-in-time because when this happens they "Punt" to SIMPLE recovery mode. I am a DBA, not a sysadmin and this drives me crazy...but I have other things to work on.
I'm sorry for your experience with VMware's support team. My current 5.1 envrionment is not exceedingly complex, but I do have several of the vCenter virtual appliances implemented, external monitoring from Solarwinds, and backups occuring, and I do not see the same transaction log bloat. Our DBA set our tlog backups to every 3 hours.
I still bet there is something configured in your environment that is causing these transactions (correctly, or not), but if your VMware admin team is OK with the current workaround, it probably isn't worth chasing down. There there has to be a reason VMware as a rule recommends/encourages simple recovery mode on the vC database, but they should elaborate on that instead of just trying hiding behind the fact that they recommend simple mode (IMO) when someone calls for support.