Neal Matthews
21st April 2005, 12:44
Hello,

We are currently running a c3sch0 to c4ags0 upgrade project. Our testing is now pretty well advanced and the performance of most of the sessions appears to be improved.

However MRP is taking 12 hrs on the new server when on our existing server it is only taking 3 1/2 hours.

The specs of the two servers are as follows;

Live server 4c3 sch0 IDS 7.31 UC6 using level 1 drivers

New server 4c4 ags0 SP15 IDS 9.40 UC4 using level 2 drivers

The existing server does not have any table boosters running for mrp.

Does anybody have any ideas for how we can improve the MRP performance with our new server as I don't really want to go live with one of our core processes four times slower than it was.

Cheers
Neal Matthews
IT Support Analyst
ATY Automotive & Industrial Components (UK) Ltd.

Dikkie Dik
21st April 2005, 15:02
Best way to analyze what is goeing wrong is making a Call Graph Profile of the MRP run. For more details I liek to refer of chapter 2 of my document that is uploaded here (http://www.baanboard.com/baanboard/showthread.php?t=7665)

Share the results and we will give you some hints.

Kind regards,
Dick

dave_23
21st April 2005, 16:03
Also, for us here, it would be good to know how you did the upgrade and
what steps you've taken so far..

For example, did you update statistics on all of your tables after the upgrade?

Dave

Neal Matthews
21st April 2005, 17:08
Thanks for the feedback guys. I will try out the call graph profile.

Regarding the upgrade we've basically set up a new server with Solaris 9 and IDS 9 with ags0 SP15 installed.

We have set up a dbmap which spreads the load across the disks and also seperates the heavy hit tables. The map has seperate dbspaces for the tables and the indexes.

The migration comes in two parts from sch0 > sch1 > ags0.

We are now regularly running update statistics medium and most of the processes on the server are running ok we are just suffering from the MRP problem.

Thanks for your interest.

Cheers
Neal

dave_23
21st April 2005, 17:10
So you had to pre/post the tables into informix? or did you use informix utilities

Neal Matthews
21st April 2005, 17:15
We used Baan for moving the data.

Cheers
Neal

Neal Matthews
22nd April 2005, 10:28
Well here's the profile I must confess I'm not exactly sure what it's telling me.

Cheers
Neal

Dikkie Dik
25th April 2005, 16:54
Well here's the profile I must confess I'm not exactly sure what it's telling me.

This is the Call Graph of the report. More files should have been created after closing the session. I am looking for a file that starts with profile.849.timrp1210m000.

Kind regards,
Dick

Neal Matthews
25th April 2005, 17:07
Hello Dikkie,

This was the only profile.849 file that appeared in $BSE/tmp.

I will run the session again tonight and see what files appear.

Cheers
Neal

Neal Matthews
26th April 2005, 11:21
Hello Dikkie,

Again I only received one file see attached.

What profile variables should I be setting. Should I be setting PROF_RTIME ?

Cheers
Neal

dave_23
27th April 2005, 04:00
Hi Neal,

How about an -- -set INFPROF=0

Should create an infprof logfile in your home directory or $BSE/tmp

Dave

Neal Matthews
27th April 2005, 13:15
Hello,

Set PROF_RTIME on my last run last night and this gave me the attached output. This appears to show me my heavy queries but I'm still not quite sure where to go with this.

Cheers
Neal