Hi,

We are facing performance issue on our Production server. I feel it is the OS resources that are lacking...

Attached documents show the output of top command immediately after server restart and after bringing up the DB. I asked our OS administrator @ the swap being used, but he says its normal for OS to start using swap space :D

I also asked our DBA if we can Lock the SGA in RAM, but he said that Solaris has problems in locking Memory....:(

Do you feel there is any problem in server memory configuration? OR anything else?

Regards,
Kaustubh

You say we have a performance problem? What is the problem? What is going slow? What are the normal times and what is it now?

Regarding the swap: UNIX and Linux reserve a lot of swap space that is normal.

Remember that when you restart the database all cache is empty and thus need to load all things from disk. This is for database, OS cache and disk cache. So the start of the first sessions indeed can take longer. They fill the cache for you. If after one hour the performance is still bad, you might consider to read this thread (http://www.baanboard.com/baanboard/showthread.php?t=7665).

Hope this helps,
Dick

Hi,

Most of the time, users from Warehousing module and Finance module face the performance issue.

Mapping integration transactions: Used to take 1 hour, but now takes more than 5 hours for almost the same range (transactions processed in 2 days).

Processes like Inbounding, Receipt Confirmation, Outbounding (am not aware of these functionalities :o ) take more than 20 minutes, which used to get completed in 4-5 minutes.

I will also check the performance tracing and tuning guide (Have never used it though :o ) and see if I can apply it on our system.

The performance improvement is seen after restarting the server, but after 2-3 days, the performance degrades...

Regards,
Kaustubh

While u troubleshoot on ur own , it may be a good case to ask OPS to investigate this issue .

What does OPS stand for?

Infor One point Support (OPS).

Hi,

We have already told them. They are looking into it.

Thanks and Regards,
Kaustubh

Hi,

We found that the OS block size is "8K", while Oracle block size is "4K". This could be the cause of performance problem (high Disk I/O).

We are working closely with Infor to resolve this issue. We might re-create the DB to change block size.

Any comments are welcome.

Regards,
Kaustubh

Hi,

We are taking adavantage of Holy Week :) down time to recreate our DB with 8K block size. We will also run the OPD (One Point Diagnostics) Tool to tune Baan/DB/OS Kernel parameters.

Regards,
Kaustubh