My question is about index usage when bitwise operations are invoked.
Lets suppose we have 2 tables TBL1 and TBL2 as the following:
integer categoryGroup; // categoryGroup is declared as an index on TABL1
integer categoryGroup; // categoryGroup is declared as an index on TABL2
By conception, I suppose that:
- [categoryGroup] may hold a limited number of values, less than 32 values.
- [categoryGroup] is of type integer => it means 4 bytes => 32 bits
=> 32 places available to hold binary '0' or binary '1' values.
- [categoryGroup] is the result of an "OR bitwise operation" among a
predefined set of variables [variableCategory].
We suppose that [variableCategory] is of type integer (=>32 bits)
and each binary value of [variableCategory] may only hold a single binary
Ex: variableCategory1 = 00000000000000000000000000000010
variableCategory2 = 00000000000000000000000000100000
variableCategory3 = 00000000000000000000000000001000
If [categoryGroup] = variableCategory1 | variableCategory2 |
=>[categoryGroup] = 00000000000000000000000000101010
I have an SQL request similar to:
SELECT ..... FROM TBL1, TBL2 WHERE
<inner join between TBL1 and TBL2 is True> AND
TBL1.CATEGORY & TBL2.CATEGORY <> 0 //-- where & is the AND bitwise
1/ IS the above SQL request will use the INDEX [categoryGroup] defined on
TBL1 and TBL2 ?
2/ What should I do or How should I modify my SQL request in order
to force the query engine to use an index ? (the already defined index or
another useful index)
Thx a lot
pgsql-performance by date
|Next:||From: Patrice Castet||Date: 2007-09-13 13:03:54|
|Subject: Clustered tables improves perfs ?|
|Previous:||From: El-Lotso||Date: 2007-09-13 07:20:13|
|Subject: Re: [Again] Postgres performance problem|