Chuck wrote:
> I have an update that I'm trying to optimize.
>
> UPDATE PS_TL_PAYABLE_TIME
> SET PAYABLE_STATUS = 'RP'
> WHERE PAY_SYSTEM = 'NA'
> AND PAYABLE_STATUS NOT IN ('PD', 'DL')
> AND PAYROLL_REQ_NUM <> 1
> AND EXISTS (
> SELECT 'X'
> FROM PS_TL_XREF_TBL A, PS_PY_XREF_WRK B
> WHERE A.PAY_SYSTEM = 'NA'
> AND A.SEQ_NBR = PS_TL_PAYABLE_TIME.SEQ_NBR
> AND B.XREF_NUM = A.XREF_NUM
> AND B.PROCESS_INSTANCE = :1)
Before anyone suggests just adding indexes to PS_TL_PAYABLE_TIME, I've
already done that. The cardinality was too low and the index got
ignored. Besides, what I'm really trying to figure out now is not how to
optimize the SQL, but why two statements with identical execution plans
running against the same data on a static test database don't do the
same number of consistent gets or run in the same amount of time.
--
To reply by email remove "_nospam"
Received on Tue Dec 06 2005 - 11:32:27 CST
