Re: Oracle vs DB2 on heavy OLTP loads

From: Serge Rielau <srielau_at_ca.ibm.com>
Date: Wed, 18 May 2005 10:01:55 -0400
Message-ID: <3f106lF51bdvU1@individual.net>

dbguy456_at_yahoo.com wrote:

>>cause of database instability is incompetent SysAdmins and DBAs

>
> That's what support wants us to think. I have quite different point of
> view though - no incompetence on the side of SysAdmin/DBA should be an
> excuse for DB instance crash, period. Poor performance? Sure.
> Invalid results? Could be.

Don't agree. Wrong results are even worse than a crash. At least a crash can't be missed.
> Diagnostic log full of error messages? Sure.
> Instance crash? No way.
>

Agreed. No ifs no buts.

Let me make a couple of simple comments. As you see from the posts the standard recommendation is O9R2 which is a stable release (no function being added for a long time).
It's noted that O10gR2 requires "guts" and I'm sure Mark T. can give you the current level of penetration of O10g. DB2 V7.2 matches O9R2 in terms of release cycle (one backlevel) DB2 V8.2 matches O10gR2 (latest drop).
The same shared customers (I know it first hand from briefings) who don't want to go to DB2 V8 for LUW also don't want to go to O10g. IBM's biggest mistake is/was to push customers to the current release by announcing end of regular support (You can run V7 supported and many customers do!) before Vnext shipped (unlike what Oracle did with O8i). You state you are affraid that DB2 V8 may not be stable before Vnext ships and then the cycle starts again.
If IBM yanks support for DB2 V8.2 before V8.2 + 2 is shipped and Vnext is stabilized similarly as V7.2 is stabalized, that would mean they/we have learned nothing.
That would be very poor indeed and I'd reconsider my employment options myself should that happen.

DB2 for LUW has grown a lot with V8. As part of this IBM needs to grow up its delivery strategy and I'm confident they will. After all DB2 V7 for zOS is supported and will be for a long time. No magic trick to it.

Cheers
Serge

PS: 24x7x365? Talk of wrong results .... ;-)

-- 
Serge Rielau
DB2 SQL Compiler Development
IBM Toronto Lab

Received on Wed May 18 2005 - 09:01:55 CDT