Oracle redo

vincem

Member
We are running Order to Cash with stub Financals under One World Xe with SP 18.1 on an AIX 4.3.3/Oracle 8.1.7.3 platform.

Recently, I was in a meeting with an Oracle DBA who expressed supprise that we were generateing 419MB of Oracle redo every 36 minutes (on average). What is your opinion? How much redo is your JDE system generating?
 
WOW,
400 MB every 1/2 hour?! Almost 20GB in 24 hours?

And you're only using a small part of OneWorld?

We generate a little over 2GB a day in archived redo logs. We have approx 100 users and use full manufacturing / inventory / sales/ procurement / financials.

Do you have any custom UBEs that use a "work" table? Something is grossly inefficient here unless you have 1000+ users.
 
1GB to 2GB a day archived logfiles for me (about 40 users). Yours is tooooo high.
 
We do approx 40 GB of redo in 24 hours. MRP/sales/financials/inv with approx 300 users. Has anyone found some way in the app to keep it from writing out so much? We now have almost 200GB of JDE/ORACLE data while running this app for 8 months. I know part of the problem is that the apps side does not have all the purge pgms in place.

Please point me to any resources (book, online, intergrators, etc) that would know how to deal with this.

Thanks in advance!
 
Kbara,

1. This should have been started as a new thread.
2. What is your system configuration info?
3. 40 GB of redo logs is a LOT of transactions. Can you give some feel as the type and number of business transactions taking place?
4. Your DBA should setup some scripts to track table growth week - to week to make sure you don't have some runaway tables/processes growing insanely.
5. If you've only been running live for 8 months you shouldn't have any purgable data yet - unless there's stuff you haven't told us (interfaces, customizations, ...)
 
One thing I know is different is that each of our branches tends to run independently with respect to MRP/MPS. So instead of running MRP and MPS once for the whole company, we tend to run 50 MRP requests and 50 MPS requests. This then gets magnified by our short business cycle (24 hours or less). We are constantly taking orders that might ship within the next 24 hours so our planners run MRP and MPS more often than once daily to make sure they always have current info. I would say most M2D planners run MPS 3-6 times daily with a median of 4 times.

Our DB has over 180GB of data in 8 months of use. Performance problems are happening although stats like buffer hit ratio are approx. 94%. We have buffer busy waits quite a bit on reads where one sql waits for the other to get done.

init.ora attached.

Sorry about not starting a new thread, to me this is/was related to oracle redo.
 
Back
Top