August 21, 2001 at 4:23 am
Hi all
At present I have both read ahd write cache enabled on my RAID-5 array (4x15k 18gig drives). From what I understand, the write-caching at the hardware level is basically hidden from SQLServer, therefore, sql*server will assume that the write was successful. This of course means without a good battery backup scheme and possibly the ability to shutdown sqlserver if the power is lost and the batteries are about to fail.
As for read-cache, I have failed to read anywhere where its a "bad" thing as such and any tips re read cache.
Any performance/best-practice information around these realms would be great.
Cheers
Chris
Chris Kempster
www.chriskempster.com
Author of "SQL Server Backup, Recovery & Troubleshooting"
Author of "SQL Server 2k for the Oracle DBA"
August 21, 2001 at 4:48 am
Nothing wrong with battery backed cache. Although tuning it is a royal pain. You've got the cache caching, SQL caching - you spend a lot of time trying to start clean for each test. Having it to start with is the big thing. I'd suggest a SWAG based on what you think your overall read/write percentage is - sorry thats not very scientific!
Andy
August 21, 2001 at 9:22 am
Well from my days with Dell, the company line was to turn off write caching to the possiblility of loosing info. Also tuning the caching is a real pain. I would turn it off run some test then turn it on and run some test. You can't go wrong with all write caching if your raid controller supports it, as a rule of thumb you will have more reads than writes. Just my .02$ worth 🙂
Wes
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply