| |
|
RSS Feed - WebSphere MQ Support
|
RSS Feed - Message Broker Support
|
  |
|
| MB to DB2 performance on z/OS |
« View previous topic :: View next topic » |
| Author |
Message
|
| Jakob_CPH |
 Posted: Mon Dec 10, 2007 7:11 am Post subject: MB to DB2 performance on z/OS |
 |
|
Apprentice
Joined: 28 Jan 2005
Posts: 26
|
Hi,
I have received a question regarding the performance of DB2 inserts from MB. The client consider to move a DB2 insert from CICS to MB on z/OS.
If you receie a message from a MQ queue and wants to put it in a local DB2 database what is then the best performing method?
Use MB more cpu to perform an insert than if the insert is done by CICS ? How much is the difference? and why?
Thanks 
Jakob |
|
| Back to top |
|
 |
| Vitor |
| Posted: Mon Dec 10, 2007 7:47 am Post subject: Re: MB to DB2 performance on z/OS |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| Jakob_CPH wrote: |
The client consider to move a DB2 insert from CICS to MB on z/OS.
|
Presumably you mean move the insert from a CICS task to MB?
I think you'd see a bigger hit from the relative performance of the task and the flow than the architecture, i.e. an optimised CICS task will outperform a badly written flow and a well written flow will outperform a badly written task.
A better question is what else is MB going to be doing? How does it fit into your general architecture? Will you gain overall savings by replacing a task with a flow?
In general an insert is an insert is an insert. Obviously your database has been properly tuned and is performing at it's maximum....?
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| Jakob_CPH |
| Posted: Mon Dec 10, 2007 1:06 pm Post subject: |
|
|
Apprentice
Joined: 28 Jan 2005
Posts: 26
|
Thanks for your reply 
My reason for asking is that I believe to have heard that MB always access a DB via a ODBC connection and that kind of connection performs in general worse than a 'direct CICS' connection to DB2. Is that just something I made up myself 
The plan is to replace a CICS task with a message flow for architecture reasons but if the MB DB insert/ lookup perform mcuh worse than a CICS DB2 insert/ lookup then we will consider other options.
What I hear you say is that a well written MB flow perform as good as a well written CICS task?
Thanks  |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Mon Dec 10, 2007 5:00 pm Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
There's always been a lot of misinformation spread back and forth about how performant it is to use Broker to do database inserts.
Some of this comes from product positioning, some of it is probably holdovers from the v2.1 days.
I don't know, myself, how comparitively good or bad Broker is at database inserts. I've never had any performance concerns doing so, though.
And to be clear, the product itself as a whole has gotten substantially faster in every release.
I would a) run tests yourself, b) review the performance spac carefully, c) *start* with your *requirements*.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| Vitor |
| Posted: Tue Dec 11, 2007 3:11 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| Jakob_CPH wrote: |
| My reason for asking is that I believe to have heard that MB always access a DB via a ODBC connection and that kind of connection performs in general worse than a 'direct CICS' connection to DB2. Is that just something I made up myself |
Kinda!
MB does use an ODBC connection. It probably does perform worse that a CICS task on a straight "how long does an insert take" but I think you'd need a fairly accurate stopwatch to detect the difference with a modern ODBC.
I also stand by my previous comment. The task isn't doing a straight insert, but an insert wrapped in a load of application logic. If the task is badly written a well written flow will leave it for dead. Even if it's well written, an equally well written flow will match it for performance in general terms.
| Jakob_CPH wrote: |
| The plan is to replace a CICS task with a message flow for architecture reasons but if the MB DB insert/ lookup perform mcuh worse than a CICS DB2 insert/ lookup then we will consider other options. |
If a flow is the correct architecture choice then you should use a flow. You'll pay more, in many different ways, and for the life of the solution, if you try and implement an non-optimal solution just because a CICS task is a few cycles faster at database work.
Another important question is how fast you need the inserts to be. For instance, if the CICS task can insert 30 rows a second, the flow only inserts 15 rows a second but the maximum throughput of the business task is 5 rows a second then you're worrying about a problem that isn't actually there!
(Timings used for illustrative purposes only. On z/OS it should all go much faster than that)
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
|
|
|
| |
|
Page 1 of 1 |
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
|
|
