Poor performance in WPD env on AS400

[Kieran]

Hi all

The users have noticed very slow performance in the WPD7333 when inquiring on data (eg Supplier Ledger inquiry), which got worse a month or 2 ago. When I run the same query in PD7333 it returns the data about 10-20 times faster than WPD7333 env. Also if I run the same queries in WPY7333 on the same AS400, the performance is also good.

There are also other screens which are slow so I'm pretty sure its not limited to a index/app, and I checked the debug log and the SQL statements are running quickly.

I tried the tests all from the same citrix server and with the same JDE login. I ran the same queries on the test AS400 in WPY7333 and they ran fine there as well. Both AS400 have the same PTF levels.

I checked the OCMs for WPD and they match the WPY

Any sugggestions on what could it be or what i should check? SQL packages perhaps? I also dont know if they are rebuilding the indexes regularly.

Thanks
Kieran

We are on Xe SP23_U1 and the AS400 is on V5R4. The citrix servers are all Win2003 R2 SP2.

 

[Devki]

Try deleting the User Overrides - only for Grid Format type. You can try deleting for 1 user and checking the response after he/she logs back on. Note that the user must log out before you delete the Grid Format in User Overrides. This has worked for me.
devki

 

[-Apple-]

Tune the performance on AS400 has many parameters
1) INI on AS400
2) Memory Management (share pool)
3) Jobque assignment (Single Thread and Multi Thread)
4) Other maintenace like Purging for PDFs, other unwanted Data for JDE
routine tables (Not production tables)
5) Job Monitoring etc.

I have just brief you on above topics.




--
Regards,
Nitin Shetye

 

[SQLGAL]

Hi Kieran -

The query optimizer on the i considers a great variety of environmental factors when making its costing decisions and building the access plan. Chances are some of them are different across all of your installed instances.

SQL indexes are of course the best source of statistics and offer an alternate query implementation method as well. Building the perfect indexes gives you the biggest bang for the buck.

As for rebuilding them regularly, there is no good reason to ever do that. It’s a ‘set it and forget it’ type of action.

 

[Kieran]

Thanks for the thoughts. We will try building a new full PD package first as the only difference I can see between WPD and PD, is where the BSFNs run.

I would have thought (correct me if I'm wrong) that the queries were run using the same indexes whether or not they are run for the client or server.

The annoying thing with this issue is that the same applications perform well in the PD environment.

Thanks
Kieran

 

[Tom_Davidson]

Kieren,

It could be your business function mappings (OCM). If you are moving a large amount of data over your network, this can take a significant amount of time.

Are you using the default OCM's for a W environment?

This is the opposite problem but for example we were using supply and demand, in the PD enviornment it took about 60 minutes to get an answer. By moving the two supply and demand business functions back to the server, and moving data at bus speed it went from 60 min to less than 3.

Just another thing to look at.

Tom