Please start any new threads on our new
site at https://forums.sqlteam.com. We've got lots of great SQL Server
experts to answer whatever question you can come up with.
| Author |
Topic |
|
Lijo Cheeran Joseph
Posting Yak Master
123 Posts |
 Posted - 2010-12-09 : 10:00:46
|
| Hi Team,While trying to tune a database, we ran a load test and the trace was passed to the DETA. The DETA returned index creation suggestions and provided 60% improvement forecast. We applied all the indexes (though it is not best practise). Still there is no reduction in Full Scans/ Sec.Can you please tell the possible reason of such a behavior? ThanksLijo Cheeran Joseph |
|
|
nigelrivett
Master Smack Fu Yak Hacker
3385 Posts |
Posted - 2010-12-09 : 10:09:07
|
| I would consider it a bad thing to do - but then I expect people to be able to optimise systems.Look at what is causing the issues and deal with that individually.First that means deciding what is an issue if anything.Making something more efficient may actually be bad for your system - I would look at what you have done so far and decide whether it is necessary or going to cause issues - or give enough benefit for the extra resources.==========================================Cursors are useful if you don't know sql.SSIS can be used in a similar way.Beer is not cold and it isn't fizzy. |
 |
|
|
Kristen
Test
22859 Posts |
Posted - 2010-12-09 : 10:50:21
|
| "or give enough benefit for the extra resources."From another thread I think the O/P is trying to meet client's specification of only 1 Full Scan per 1,000 Index Requests (or something like that) |
|
|
|
GilaMonster
Master Smack Fu Yak Hacker
4507 Posts |
Posted - 2010-12-09 : 11:41:32
|
| That's a stupid specification. Maybe I'm old-fashioned, but I believe in tuning queries for performance, not for some arb number for a perfmon counter.So 1 table scan on a 10 000 000 million row is fine, but 2 scans on a 5 row table isn't?--Gail ShawSQL Server MVP |
|
|
|
Kristen
Test
22859 Posts |
Posted - 2010-12-09 : 12:29:48
|
| Yeah, I suggested in the O/Ps other thread that "Response time" - e.g. "Task-X should take < 2 seconds" would be a better metric. Worrisome that the O/P's client has imposed an Acceptance Test using such a unit IMHO |
|
|
|
|
|
|
Database Engine Tuning Advisor Question