What we do is:
Pro*C batches extract the desired data on the lowest levels in the various
time, product and organisation hierarchies. Also changes in the hierarchy
(reclassifications) are extracted. This is written in flat files.
The flatfiles are ftp'ed to the DWH server.
Here the flatfiles are SQL*Loaded into the atomic tables.
Other Pro*C batches are run for generating syntethic keys and populating the
datamarts, and also pushing down the Majro Changes in the hierarchy..
A combination of Pro*C batches and SQL-scripts are run for
generating/calculating various aggregated DataMarts.
All this is fully restart/recovery enabled (including the flatfile writes)
and monitorable from our Redwood Scheduler.
We have sometimes tried to do this over database links (queries on the DWH
server collecting the Atomic data) but were not content with reliability and
restartability.
Good Luck,
Erik Ykema
"Kimo" <ora_kimo_at_yahoo.de> wrote in message
news:6173a20e.0301210756.52fbc275_at_posting.google.com...
> Hi....
>
> In order to extract, transfer and load to our Data Warehouse we have
> created 2 concepts,
> 1- export the needed data to temporary tables on the live database and
> then transfer and load the data into Data Warehouse.
> 2- export the data to temporary tables on a separated database and
> then transfer/load the data into DWH.
>
> we concept would you recommend me and can anyone tell me the
> advantages and
> disadvantages of each concept ?
>
> You can also tell me if you have another option !
>
> thanx
> kimo
Received on Sat Jan 25 2003 - 13:51:58 CST
