I would like to suggest review some points before you start to
online and batch concurrently.
-If you are on V4 or later , use index type 2 .
- Plan or Package definitions. If you want to improve your
have bind your aplications chosing the options that can help you to
like ISOLATION (CS) , ACQUIRE(USE), and RELEASE(COMMIT).
Isolation level CURSOR STABILITY allow your application lock the
enough meantime the cursor is getting another page. Acquire(use)
for packages) and Release(commit) allow your application free
frecuently but it depend on the commit frecuency.This combination
the greatest concurrency.
-Commits Fecuently, allow your aplications avoid the amount of
-Try to set up a bufferpools big enough (if you can afford it) to
a big number of the getpages request ( high hit bpool ratio).
-One important thing that you have to check is the update activity
db2 subsystem. If it is high to have to specify a big OUTBUFFER to
the updates occurred on your system and set WRTHRSH good enough to
OUTBUFFER available to news updates.
And the LOGLOAD has to be checked too .
Those are some points and I hope it helps you.
>From: [login to unmask email]
>Reply-To: DB2 Data Base Discussion List
<[login to unmask email]>
>To: [login to unmask email]
>Subject: Implications of running batch and online
>Date: Thu, 6 Jan 2000 10:45:26 -0500
>We are a shop which runs CICS transactions during the day and
batch jobs at
>night after online update has been programmatically disabled.
>investigating the possibilities of running batch and online
> I am interested in methods used to allow this. Our batch
>currently take table locks and commit at the end of processing.
>tables are defined with page-level locks. What sort of
>programmatic(locking, commits, rollbacks, restarts,
>considerations on read/update) and/or database changes would be
Get Your Private, Free Email at http://www.hotmail.com