| Author |
Message
|
| happyj |
 Posted: Thu Jun 01, 2006 12:31 am Post subject: compute or database node ? |
 |
|
Voyager
Joined: 07 Feb 2005
Posts: 87
|
Are there any advantages in using
a database node instead of a compute node
for a simple select/update db operation?
WBIMB V5 CSD6 / Oracle DB
many thanks |
|
| Back to top |
|
 |
| pottas |
| Posted: Thu Jun 01, 2006 1:03 am Post subject: |
|
|
Disciple
Joined: 27 Oct 2005
Posts: 185
Location: South Africa
|
Hi happyj,
The main differences:
If you want make changes to your output message, you have to use a Compute Node, you cannot manipulate your message in a Database Node. Of course you can still make updates to a database from a Compute Node.
If you are only going to make updates to the database based on the incoming message data, you can use a Database Node.
...so, to my opinion, based on the above, you then have to decide which are going to serve your purpose. If it is my call - I always use Compute Nodes in my flows...
pottas |
|
| Back to top |
|
|
| happyj |
| Posted: Thu Jun 01, 2006 1:12 am Post subject: |
|
|
Voyager
Joined: 07 Feb 2005
Posts: 87
|
pottas
thanks for your response.
In this example I don't need to change the message tree
so I can use either.
Are there any performance improvements in using the database node ? |
|
| Back to top |
|
|
| pottas |
| Posted: Thu Jun 01, 2006 2:09 am Post subject: |
|
|
Disciple
Joined: 27 Oct 2005
Posts: 185
Location: South Africa
|
Happyj,
Well, I had a look at the performance report for WBI and it is not too clear which one is the fastest, my opinion, there's not much difference performance-wise (maybe someone on this forum can comment on this). Even though the two nodes can be used for the same purpose (i.e. Database I/O), they are different in some respects as mentioned.
We here at our site uses Compute Nodes for everything, regardless if we modify the message or not. The reason being, if in future you need to modify your OutputRoot, you don't want to be stuck with a Database Node.
If you want to have a look at the performance reports:
http://www-128.ibm.com/developerworks/websphere/library/techarticles/0403_dunn/0403_dunn.html
...and for version 6:
http://www-1.ibm.com/support/docview.wss?uid=swg27007159
Hope this helps.
Good Luck!!
pottas |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Thu Jun 01, 2006 5:14 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
I guess at the simplest level, the database node will be slightly faster simply because it doesn't have to copy the message tree for editing.
I don't think it will be noticeably faster because of this, except for very large messages.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| jbanoop |
| Posted: Sat Jun 03, 2006 4:50 am Post subject: |
|
|
Chevalier
Joined: 17 Sep 2005
Posts: 401
Location: SC
|
If there is no appreciable performance gain with a database node then why provide a different node alltogether ? I mean wouldnt it be easier for everyone to just use compute node (I guess most already use it in stad of a database node) rahter than go for a database node which does not add much value in the end??
Anoop |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Sat Jun 03, 2006 6:58 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
| jbanoop wrote: |
| If there is no appreciable performance gain with a database node then why provide a different node alltogether ? |
In the 2.1 days, the instructor for the training I took joked that the development team got bonuses for each node they built.
I'm sure this is not true.
I might be wrong about the performance differences, too.
But there's at least some value in being able to look at a node and know that all it does is interact with a database - and not force you to look at the ESQL.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
|
|
