Icon View Thread

The following is the text of the current message along with any replies.
Messages 1 to 3 of 3 total
Thread Commit count
Mon, Dec 17 2007 3:13 PMPermanent Link

Allan Brocklehurst
Hello;
Is there a formula I can use to determine the optimum Number to force a commit;  Ie record length, memory etc, I don’t know.

What I did:
I Fired up the code with a

(A)
START TRANSACTION

I processed all of my records (358K) then
Forced a COMMIT

This worked very well. It slowed on the actual write to disk

(B)
I then did a commit after every 100 records. This was unbearable slow.

Then

I noticed that in (A) the system smoked thru 30K before slowing down
I then re-did (B) committing every 30K records this worked great.

Thanks in Advance for any direction

Allan
Tue, Dec 18 2007 3:06 PMPermanent Link

Tim Young [Elevate Software]

Elevate Software, Inc.

Avatar

Email timyoung@elevatesoft.com

Allan,

<< Is there a formula I can use to determine the optimum Number to force a
commit;  Ie record length, memory etc, I don’t know. >>

It really is a matter of playing around with the commit interval.  The
reason for this is a combination of the system load, hard drive speed, row
size, and what the buffering settings are for the TDBISAMEngine component.
In general, the larger the row size, the lower the commit interval, and
vice-versa.

--
Tim Young
Elevate Software
www.elevatesoft.com

Wed, Dec 19 2007 7:48 AMPermanent Link

Allan Brocklehurst

Thanks Tim.

"Tim Young [Elevate Software]" <timyoung@elevatesoft.com> wrote:

Allan,

<< Is there a formula I can use to determine the optimum Number to force a
commit;  Ie record length, memory etc, I don’t know. >>

It really is a matter of playing around with the commit interval.  The
reason for this is a combination of the system load, hard drive speed, row
size, and what the buffering settings are for the TDBISAMEngine component.
In general, the larger the row size, the lower the commit interval, and
vice-versa.

--
Tim Young
Elevate Software
www.elevatesoft.com

Image