>>>>I'll try later - busy with work right now. I tried on the tables I had in my test database (in SQL Server).
>>>
>>>Ok I just tried for you:
>>>
>> t6 t5 t4 t3 t2 t1
>>>Client processing time 0 0 0 0 0 16
>>>Total execution time 2328 2296 3046 2250 2328 5937
>>>Wait time on server replies 2328 2296 3046 2250 2328 5921
>>>
>>>
>>>Two passes. t1,4 (LEFT), t2,5 (NOT IN), t3, 6 (Exists).
>>>Cetin
>>
>>In your tests performance of the LEFT JOIN seems to be much worse than others. What are tables metrics and how many records did you get as a result of a query? Also, did you discharge the results or not?
>
>Yes as I told you before it was the slowest. Metric is simple. 2 million records table created, and copied onto another (structure was as simple as it can be, an ID plus a varchar data field. With a more complex structure it is likely it would be much slower). Then a few more records are added to one of them. Index created on join field.
>Cetin
Just want to share this link
http://sqlinthewild.co.za/index.php/2010/04/27/in-exists-and-join-a-roundup/ that contains performance results for
IN/EXISTS/INNER JOIN and NOT IN/NOT EXISTS/LEFT JOIN by Gail Shaw.
If it's not broken, fix it until it is.
My Blog