| Author |
Message
|
| kutulu2000 |
 Posted: Tue Sep 05, 2006 9:18 am Post subject: Different Aplications Running in the same instance of MQWF |
 |
|
Acolyte
Joined: 08 Nov 2005
Posts: 63
Location: Mexico
|
Hi,
Actually we have two different applications running each one on a different instance of MQWorkflow, each application has different architecture and services to provide. I think that these applications can't coexist in the same instance of MQWorkflow, because the data structures are differents ( one uses only the Default Data structure full of information and the other uses lightweight data structures ) , one these applications uses triggers on the FMC.WORK_ITEM table to insert information into another data base, etc, but the people who take the decisions about the budget want to these applications run over the same MQWF server, sharing the same database, resources, etc.
Has anyone had any experience in a situation like this?
Can these applications run together?
_________________
#10 |
|
| Back to top |
|
 |
| Ratan |
| Posted: Tue Sep 05, 2006 12:03 pm Post subject: |
|
|
Grand Master
Joined: 18 Jul 2002
Posts: 1245
|
I dont see any reason why they should not run on the same MQWF Instance unless there are any performance reasons to seperate those.
_________________
-Ratan |
|
| Back to top |
|
|
| kutulu2000 |
| Posted: Tue Sep 05, 2006 1:19 pm Post subject: |
|
|
Acolyte
Joined: 08 Nov 2005
Posts: 63
Location: Mexico
|
Both applications must implement searches into the same WORK_ITEM table, and one of them performs these searches over a 30 data members using LIKE and the other performs the search over a lightweight data structures (7 or 8 data members). One of them is managed by another Fat Client Application ( poor implementation for a Workflow Concept ) and the other is a Thin Client managed by MQWorkflow itself. Obviusly, you must have the complete information to make a decision, but I want to know your opinion with these elements.
Thanx
_________________
#10 |
|
| Back to top |
|
|
| Ratan |
| Posted: Wed Sep 06, 2006 3:43 pm Post subject: |
|
|
Grand Master
Joined: 18 Jul 2002
Posts: 1245
|
Type of Client does not matter.
_________________
-Ratan |
|
| Back to top |
|
|
| jmac |
| Posted: Thu Sep 07, 2006 7:21 am Post subject: |
|
|
Jedi Knight
Joined: 27 Jun 2001
Posts: 3081
Location: EmeriCon, LLC
|
| kutulu2000 wrote: |
| Both applications must implement searches into the same WORK_ITEM table, and one of them performs these searches over a 30 data members using LIKE and the other performs the search over a lightweight data structures (7 or 8 data members). |
Are you saying that you have code which is querying the table? Or that you are using the API to query the table. Sounds to me like the former. This is not a good idea, and would likely not be supported.
_________________
John McDonald
RETIRED |
|
| Back to top |
|
|
| kutulu2000 |
| Posted: Fri Sep 08, 2006 3:49 am Post subject: |
|
|
Acolyte
Joined: 08 Nov 2005
Posts: 63
Location: Mexico
|
No no, both applications uses API to query the table but one of them uses the default data structure full of information and access to the data members by puting on the description field of each program activity using %data memeber1%data member2%....%data member30%. The other one uses several light weight data structures and access to the data members only putting key data members on the description field. I think that if both applications searches on the same WORK_ITEM table, the performance will be affected in a very important way.
_________________
#10 |
|
| Back to top |
|
|
| jmac |
| Posted: Fri Sep 08, 2006 5:28 am Post subject: |
|
|
Jedi Knight
Joined: 27 Jun 2001
Posts: 3081
Location: EmeriCon, LLC
|
GOOD... What you might consider is putting an index on the table to help the query. Use of LIKE will always add a db performance burden, but with a proper index it should be OK.
_________________
John McDonald
RETIRED |
|
| Back to top |
|
|
| mzg |
| Posted: Thu Sep 28, 2006 5:49 am Post subject: |
|
|
Newbie
Joined: 01 Oct 2003
Posts: 8
|
Creating an index on the description will definitively help, but not in the case where you do a filter with
description like '*text*'
Here you'll have a nonindexable predicate which will result in a table scan (or at least an index scan).
Description like '
will fully utilize the index.
Good Luck!
_________________
Cheers,
Richard.
-----------------------
IBM Websphere MQ Workflow & Process Choreographer Performance Focal Point |
|
| Back to top |
|
|
|
|
