June 8, 2007 at 12:41 am
This one goes out to all the gurus.
We are in the process of migrating our entire company to a new software package.
Given my hardware and databases outlined below what would be the best file/File group and disk configuration for my initial setup.
Since this is a new install I do not have any historical performance data to base the decsion on.
So once in production adjustments will need to be made.
Hardware:
I have two Ibm Blade servers that are set up as a fault tolerant cluster.
I also have one IBM Blade server that will be a dedicated test server.
each blade has dual 2.4s with 8gb of ram. the os and sql server are installed on local disks
These are all in the same chassis
They are connected via two fiber channels to an IBM san.
I have a total of 10 146 GB disk in the san that I can configure any way I want.
we are running 2005 enterprise sp2
I will have two production databases that our application uses.
They application is designed around the premise that the two database be on the same server.
one database is 50GB and the other around 10gb
I would estimate no more then 30% size growth per year.
At any one point in time I will have about 200-250 users connected.
Thank you for any assistance you can provide on how I might configure this.
June 11, 2007 at 4:15 am
Hi Jason,
If it were me, I would keep it fairly simple. 2 disks in a Raid 1 configuration for your log files. The other 8 disks in a Raid 10 (mirrored then striped) configuration. You should get excellent performance from this configuration.
HTH,
John
July 6, 2007 at 12:06 am
Also you can place the tempdb ib RAID 1 disk array so that you can get better performance. Place the system and user dbs in diffferent drives.
Cheers,
Sugeshkumar Rajendran
SQL Server MVP
http://sugeshkr.blogspot.com
December 4, 2007 at 11:27 am
and having been down this road several times at several companies, the best solution is always here;
http://www.symantec.com/business/products/overview.jsp?pcid=2247&pvid=31_1
adds a bit of a cost, but it is 'rock solid' geo clustering solution and it does not care about versions, SAN or heartbeat link speed.
imagine 2 nodes, 2 SANS, 2 datacenters and a single DNS name for an app to connect to. failover takes less than 3 minutes in most cases and you can apply patches and hotfixes with 'zero' downtime.
Viewing 4 posts - 1 through 3 (of 3 total)
You must be logged in to reply to this topic. Login to reply