October 1, 2009 at 7:21 am
Fastest path to less timeouts would be to bury the problem in super hardware 😀
Short term, you aren't really going to get much bang for your buck.
Best option with the disks you have would be to seperate them into two RAID 1 arrays. Put OS/Paging File, Log files and backups on one array and the data files on the other array. Bear in mind that means a full server rebuild :crazy:
I wouldn't expect an external disk (presume you're thinking USB hard disk) to help much either. You will still be incurring the I/Os reading data from the datafile.
Splitting the database would just shift the I/O from one file to another on the same disk.
Hard yards in query optimisation and index tuning will be the least disruptive, but you'll probably have to live with timeouts for a little longer.
--------------------
Colt 45 - the original point and click interface
October 1, 2009 at 7:08 pm
all good idea Phil. I've also had this response from our datacenter provider:
<snip>
Currently, 99.9% of memory is being used by SQL mostly leaving no memory for windows. Also, your pagefile is set to "system managed".
We would like to do the following to start.
1. Set pagefile to min and max on E to min and max of 150% of 16GB ram.
2. Set Min and Max memory used in SQL to 12GB. This will leave 4GB of ram for Windows to operate.
This will require a reboot. I would recommend doing this as soon as possible.
Please advise with time (w/timezone.
</snip>
What do you think?
October 1, 2009 at 7:16 pm
Both valid points. But bear in mind that reducing SQL Server memory even further will push more work onto the disks where your real problem is.
--------------------
Colt 45 - the original point and click interface
Viewing 3 posts - 16 through 17 (of 17 total)
You must be logged in to reply to this topic. Login to reply