I agree with the last two posts. I got it wrong cause I can't mind read.

I can think of many examples where the cast is not implicit...

I agree with Grant's last statement.

There shouldn't be any look-ups if the index is covering all the columns in the sql statement. The quest is not clear enough in my...

Hi,

Are those estimated or actual plans?

It's best to supply the actual plans.

Could please also supply the graphical plans attached as .sqlplan in a zip file.

Thanks

I also see that perhaps we could build you some better indexes. As there are 10 bookmark lookups in the first query and bookmark lookups are normally very expensive.

I also...

By looking at the execution plans.

SQL has worked out that when you are not returning any columns from the accounts table. That this table is not actually need in the...

could you upload the actual execution plan for us as well?

edit:

Please also give us the definitions of the current indexes on those tables.

Hi Mel,

Was this section of code still needed?

PointsCategory = ISNULL(NULL, PointsCategory)

I thought it was decided that the catch -all approach would be performance problem and that perhaps calling a "Master"...

I had a feeling I should I have done some testing before opening my big mouth.

Thanks for clearing that up for Gail, I must have missunderstood completely.

🙂

This might be sound stupid, I'm not sure if it makes a difference at all I might need to do some testing.

but creating the clustered Index on #TopScores after the...

I agree with Jack.

Also Mel could you post the latest copy of the solution , I've lost track of where you all got to.

Also I do really really recommend going through the index posts on Gail's blog. She is brilliant when it comes to perf tuning and indexes. I learn't a lot of...

I'm happy to try and help as best I can.

Ok so the reason I select ID as the clustering key is because it seems to be used in all your...

if you remove the include from the index it will work for 2000, but then it would cover the full query for output fields

Hi ,

Ok first thing is first. I would definitly try and rewrite all that code to fit the ansi standard. This is a very old style of writing SQL and...

Ok this is always a hard one.

Some advice, I would try and key your Clustered index as narrow as possible.

Also before just creating or changing indexes, have you identified the...