cancel
Showing results for 
Search instead for 
Did you mean: 

New ODS using already in use LIS extractor

Former Member
0 Kudos

Hi,

I've created a new ODS objct that makes use of the data from 2LIS_05_Q0ITEM. This datasource already runs as a delta into a delivered cube that has been in use for several months. After noting a full load infopackage only pulls the data from the last time the setup tables were rebuilt, I read several threads on the topic of the setup tables, full loads, delta queues, etc. And I think I understand most of it, but am worried about trying to rebuild the setup tables without re-initializing the existing cubes.

Here are my questions:

In order to populate a new ODS with LIS related data that is already in existing cubes, I understand the steps to be:

1) Run all delta extracts for the related LIS area (in my case, 05) twice to clear the delta queues

2) Delete and rebuild the setup tables in SBIW.

3) Run a full load into the new ODS.

4) Allow deltas to continue running as normal with the infopackage now set to include the new ODS as well as the existing targets.

Is that a complete set of the steps?

What are the risks?

Would it be better to clear all related inits (RSA7) and run new inits without data to reset the delta extract queues?

I believe this should work for all LIS extractors set up as "Queued Delta", would it matter if they are set to one of the other 2 options?

Thanks in Advance

Keith

Accepted Solutions (0)

Answers (1)

Answers (1)

Former Member
0 Kudos

You will need to drop init and re-init - to include the new ODS (as I understand it). If so, you will have to make sure that txns are not getting posted while the delta is not established.

It wouldn't matter whether the update in direct or queued or V3 on the R/3 end, if that is what you ask.

Former Member
0 Kudos

We've loaded other ODS with full loads and then allowed the existing delta process to keep it up to date. These were not from the LIS area, but I don't believe that should not make a difference on the BW side.

I do worry that I'll have to clear the QM cubes already in place and re-initialize the whole set if I recreate the setup tables--however, other threads indicate that I would not have to.

Keith

justin_molenaur2
Contributor
0 Kudos

This is according to my understanding of LO, and how I think I would handle it in your scenario.

1. Block postings in R/3

2. Clear the delta queue via loading existing objects with the data (Run the periodic job and then the delta InfoPackages twice)

---At this point you have all the possible data in R/3, and the document block is in effect, no new data can be created relevant to BW

3. Delete setup tables

4. Fill setup tables

5. Run an init load <u>only</u> into the new objects (OR an init without data and multiple full repair loads if the data size is large) - easiest to use new InfoPackages

-- As soon as there is a successful initialization of the 05 application area into BW, the delta mechanism is functioning and will capture changes and new documents

6. Turn the document block off

7. Continue delta loads as normal to <u>all</u> objects with existing InfoPackages

Hope this helps,

Justin

Former Member
0 Kudos

Thanks Justin, it does help.

I considered the "Block postings" as a given--must be the DBA in me.

For step 5, I assume I could do a full load into the new ODS, clear the delta for the one datasource that I'm using on the new ODS and then run an init without data to reset the deltas. The full load for the one ODS is not so large as to be an issue (rough 90,000 records).

My primary concern is that I do not want to have to re-initialize all of the LIS QM datasources due to the setup table rebuild.

Keith