Hi!

Try following query:

SELECT A.F1, A.F2, B.F1 ;
FROM A INNER JOIN B ;
ON A.F1 = B.F1
WHERE B.F2 = "1234" ;
INTO CURSOR mytemp

I guess it makes no difference, but who knows.
In addition, filtered indexes are never used in the Rushmore optimization, so avoid it at all cost, include never use the 'deleted()' filter.


>>Hi,
>> I have 2 tables A, and B. Table A may contains 5000 records and B may contain more than 20,000 records (but just 3 fields - Total 30 Bytes). I have the SQL SELECT as the following:
>>
>>

>>
>>SELECT A.F1, A.F2, B.F1 ;
>>FROM A, B ;
>>WHERE B.F2 = "1234" AND A.F1 = B.F1 ;
>>INTO CURSOR mytemp
>>
>>Indexes:
>>A - F1 (Primary FILTER), F1 (Regular)
>>B - F1 (Regular), F2 (Regular)
>>
>>

>>
>>The SQL above take 15 seconds to load 5000 matched records from server trough 10BASET LAN. Is it acceptable? Is it possible to make it faster?
>>
>>I did try the following SQL, it take < 7 seconds to load 5000 from table B.
>>
>>

>>SELECT B.F1 FROM B WHERE B.F2 = "1234" INTO CURSOR mytemp
>>
>>

>>
>>What do I do wrongly for my 1st SQL?
>>
>>Thank you
>
>Looks to me like your SQL is fine. Check out the SYS(3054) function to see your query optimazation level. Also, be sure the stucture of a.f1 matches b.f1 to make sure rushmore is fully optimized.