I have an Exchange 2010 server running on a SAN-backed platform. The platform does block-level backups based on a snapshot/incremental basis, that only capture changed data. I was surprised to see a regular period of time where the data changes were happening at a high, sustained rate. Due to the way this system works, that can lead to >1.2TB of stored data per month.
The regularity implied a scheduled task, but it is not a fixed interval. It is approximately every 26-32hrs. The disks were performing read operations of ~5MB/s and write operations of ~4.5MB/s, for a period of 3-4hrs. The total written data was ~55-60GB.
Reading on TechNet, I am wondering if the following is causing this:
http://blogs.technet.com/b/exchange/archive/2011/12/14/database-maintenance-in-exchange-2010.aspx#checksumming
The somewhat restrictive thing is that the process only happens at most once every 24 hours. I was able to investigate while it was running, finding the following:
- the process is store.exe
- it is working on the mailbox database files
- while running, it is generating .log files (in the mailbox database folder) consistent with database changes
- the mailbox database is ~60GB in size, which fits with the total data changes on each iteration
I have currently switched to a fixed maintenance window, as a test. It's not clear whether this is the cause, as the symptoms fit, but are not conclusive.
Does anyone have any suggestions for additional troubleshooting?
Best Answer
I think you've answered your own question - in that technet blog, it says
"Exchange scans the database no more than once per day. This read I/O is 100 percent sequential (which makes it easy on the disk) and equates to a scanning rate of about 5 megabytes (MB)/sec on most systems."
On another note nothing to do with this question, is your Exchange mailbox server aware of this backup done by your SAN? If not, the transaction logs will not be cleared, which is very bad.