cancel
Showing results for 
Search instead for 
Did you mean: 

Data Migration Guidelines

Former Member
0 Kudos
42

Hello colleagues!

At the present, I have several ODS with millions of records (2 years back data). Because of a new version of my data model, I need to create new ODS objects, similar to my exisiting ones, not identical because several data fiels are removed and some others are new. Of course, I need to migrate data existing on my current ODS objects to new ODSes I will create.

Are there any general recommendations or guidelines to be followed during a data migration like the one I am describing.

I appreciate any comments, recommendations, guidelines, or experiences you could share with me on this regards.

Thank you in advance.

Mario.

Accepted Solutions (1)

Accepted Solutions (1)

Former Member
0 Kudos

Hi Mario and welcome on board !

Try to do as many (smaller) loads as you can (via parallel not overlapping full loads by using the infopackage selections)...clearly keep in mind that you need to have available work processes on your server (about this, ask for some support to your basis team or ask to have more work processes...).

Then avoid to flag the BEx reporting checkbox if you don't need it: this avoid a lot of effort when you will activate on your new ODS...

Hope it helps!

Bye,

Roberto

Former Member
0 Kudos

Hi Roberto,

Thanks for your answer.

What advantages do you see by doing numerous small loads instead of, let's say, "big-bang" approach? I mean, why not thoroughly uploading all ODS contents into another ODS?

Is it less probable to have memory or disk swap issues this way?

Thanks again.

Mario

Former Member
0 Kudos

Mario,

you already highlighted the possible issue...then, don't forget that if you load several small packets in parallel, first of all you can speed up your loading time and then you don't need to activate one unique huge request, but you can split the effort required in different time interval !

This approach has been already tested with success in the past...

Hope it helps!

Bye,

Roberto

Answers (0)