Subject: | |
From: | |
Reply To: | |
Date: | Tue, 3 Mar 1998 07:15:41 -0600 |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
At 01:40 PM 3/2/98 +0000, Paul H. Christidis wrote:
/snip
>While the process goes through a million plus entries I just feel that
>20+ hours is just too much. Can anyone hazard a guess?
Paul,
By chance is the process doing DBPUTs (detail or master) whereby a
master (related or the target set) has either a tight capacity and/or
perhaps has a binary-type key (I, J, etc.)? Seems from the stack
markers that the process appears to be searching a lot of file
pages, so perhaps you have a clustering problem in the master and
a lot of synonym chains to walk for each put. When it completes
traversing the synonym chain and determines the entry does not
exist, perhaps it then has a large forward serial search to perform
to find an open entry in the master to add the new entry and link it
to the synonym chain.
Just a thought.... not really enough data on the specifics of
the operation and base to know if this is a possibility or not...
/jf
_\\///_
(' o-o ')
___________________________ooOo_( )_OOoo____________________________________
Tuesday, March 3rd
Today in 1820 - The Missouri Compromise passes.
1845 - Florida became the 27th state.
1931 - "The Star-Spangled Banner" became our
national anthem.
___________________________________Oooo_____________________________________
oooO ( )
( ) ) /
\ ( (_/
\_)
|
|
|