Viewing 15 posts - 5,461 through 5,475 (of 5,843 total)
1) Using functions like this very often lead the optimizer astray. It can't perform properly because it doesn't know how many rows and what the value distributions are going...
December 11, 2007 at 6:41 am
That is just one of the many advantages to being a consultant - CTOs actually PAY me to tell them their stuff is poorly configured, improperly designed, sized, maintained, etc,...
December 10, 2007 at 1:36 pm
My comment was directed at an optimally sized/configured SAN. I don't consider a SAN with 512 LUNs to be optimally configured for database throughput. DAS with that many...
December 10, 2007 at 12:33 pm
Nice site name colin! 😀
I note, however, that your links on the 2005 page end in .mht for URL as opposed to htm. I tried to hit the...
December 10, 2007 at 12:28 pm
1) You state "started throwing this error" but don't specify an actual error.
2) Sprocs will often run differently than hard-coded statements. With manual executions the query optimizer has explicit...
December 10, 2007 at 8:52 am
A bulk insert method will beat individual inserts in most if not all cases. Here you would avoid the network and execution hits in addition. Linchi Shea did...
December 10, 2007 at 5:24 am
Hey, thanks for the flattery!! hehehe
You are correct though - seems that a LOT of people out there have misconceptions about sets and ordering.
December 9, 2007 at 1:16 pm
Stealing my thunder Gila! I already posted on that yesterday at 8:54. 🙂
December 8, 2007 at 1:48 pm
A trigger based system could work, as could an asynchronous mechanism other than replication as well.
December 8, 2007 at 8:04 am
Two points you are missing in the performance equation Colin. With a SAN it isn't just about the disk I/O capability. SANs usually come with large (sometimes many...
December 8, 2007 at 8:03 am
The big thing to take into account with NOCOUNT (or read uncommitted isolation level) is the fairly recently (last 18 months or so I believe) discovered situation whereby you can...
December 7, 2007 at 8:02 am
I 'believe' the optimizer can choose an allocation unit scan instead of a clustered index scan - which would allow unordered rows returned from your table with a clustered index....
December 7, 2007 at 7:58 am
Try running the dbo/nondbo test again where there are objects with the same name owned by the non-dbo user. See if anything is different then.
December 7, 2007 at 7:40 am
Be careful Ramesh - I don't believe that 'special' feature of sql server on the UPDATE statement is guaranteed to work in the future. It may also be dependent...
December 7, 2007 at 7:07 am
Well, with 2 drives per RAID set you don't have RAID 1+0 - simply RAID 1. You need 4 drives min for RAID 10.
Your setup is reasonable. I...
December 6, 2007 at 2:46 pm
Viewing 15 posts - 5,461 through 5,475 (of 5,843 total)