SQL Server 2005 Configuration Blog #2.doc

re: SQL Server 2005 Configuration Blog #2.doc

Leigh — Tue, 22 Nov 2005 17:56:20 GMT

Hi, please post the article or link to Deploying SQL Server 2005 with SAN #2, I have read #1 and #3, but missed #2.
Thanks.

re: SQL Server 2005 Configuration Blog #2.doc

Ira Pfeifer — Tue, 29 Nov 2005 20:16:13 GMT

This article is great - thanks for the recommendations. In my ongoing efforts to maximize our production SAN performance, I have a followup (or real-world application) question.

Given a certain size pool of disks (say 20 disks to make the math easy) from which to create LUNs for transaction logs for 10 databases on 5 different servers, what is the best way to minimize response time while maintaining adequate throughput to avoid maxing out the disks IO bandwidth? I can see several possible approaches:

1) 5 RAID-10 LUNs across all 20 disks (1 T-Log drive per server, sharing disks)
2) 10 RAID-10 LUNs across all 20 disks (1 T-Log drive per DB, sharing disks)
3) 5 RAID-10 LUNs across 4 disks each (1 drive per server, dedicated disks)
4) 10 RAID-1 LUNs across 2 disks each (1 drive per DB, dedicated disks)

Assuming all DBs are active all the time, it seems to me that option 3 would be the fastest, as the bigger LUNs would reduce latency and increase throughput over the RAID-1 pairs while only suffering from contention between 2 DB log-writers. Is this accurate? Anything to add or correct in this analysis?

re: SQL Server 2005 Configuration Blog #2.doc

orbhot — Tue, 20 May 2008 17:11:33 GMT

Leigh,

The search on this site doesn't work well, but try using the tags link on the side:

http://blogs.msdn.com/sqlcat/archive/tags/Performance+and+Scalability/default.aspx?p=2

Vince

Microsoft SQL Server Development Customer Advisory Team SQL Server | Green Tea Fat Burner

Mon, 08 Jun 2009 05:34:09 GMT

PingBack from http://greenteafatburner.info/story.php?id=3867