October 8, 2015 at 10:07 am
Hi
We have a SQL Server running on a virtual platform.
I was wondering if the old recommendations w.r.t disks still make a difference, things like
1. Logs on one disk data on another
2. Minimise autogrow events size DB appropriately
I know nothing of how virtual servers work, but I guess at the HW level its just one disk.
What recommendations for SQL config are important.
Alot of what I read goes over my head
October 8, 2015 at 10:17 am
terry999 (10/8/2015)
HiWe have a SQL Server running on a virtual platform.
I was wondering if the old recommendations w.r.t disks still make a difference, things like
1. Logs on one disk data on another
2. Minimise autogrow events size DB appropriately
I know nothing of how virtual servers work, but I guess at the HW level its just one disk.
What recommendations for SQL config are important.
Alot of what I read goes over my head
Depends how your virtual machine is configured (single virtual hard disk, multiple virtual hard disks, raw device mappings)
It also depends how these virtual hard disks sit within the virtual infrastructure, i.e. in the same datastore.
Always size the databases appropriately whether on VM or not.
Which hypervisor are you using?
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
October 8, 2015 at 10:38 am
If your virtual disks are sourced from a SAN and all your virtual disks come from the same disk group within the same storage, then there shouldn't be much of a performance difference splitting up the disks as all the data is spread across all the disks in the disk group. There may be some performance issue with contention for disk access in OS or the hypervisor but if there is any it might not even be perceptible. (ie just a theory: on a four core machine you could access four disk files simultaneously)
Some storage allows for specific data to be locked into higher performing tiers (SSD/15k) which would be needed if you wanted tempdb to be on SSD and data to be on 15k. The downside of doing something like this would be that this part of storage would not be available to the rest of the machines accessing the storage.
I still split up my drives for for a couple reasons. One, if the need arises to lock some data to a higher tier within the storage, no changed would have to be made to the SQL instance/database, the change would occur within the SAN. Two, if there is OS or hypervisor disk corruption and everything is on one disk all is lost upto your last backup, assuming it wasnt on the same disk as well. If data and logs are split and one of those disks is corrupted, you could minimize data loss with having just data or log+backup available.
October 8, 2015 at 10:39 am
Thanks for replying Perry.
I have no idea how it is configured as Virtual server e.g. on the physical machine.
From user of the virtual machine it appears at 2 disks, one for OS and one for Data.
October 9, 2015 at 2:23 am
These are good questions about IO but don't forget to think about CPU and memory allocation too.
p.s. don't forget to do some sort of pre-deployment IO testing.
October 9, 2015 at 7:39 am
terry999 (10/8/2015)
Thanks for replying Perry.I have no idea how it is configured as Virtual server e.g. on the physical machine.
From user of the virtual machine it appears at 2 disks, one for OS and one for Data.
You'll need to contact your admins and get the info for us to help further
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
Viewing 6 posts - 1 through 5 (of 5 total)
You must be logged in to reply to this topic. Login to reply