Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Usenet -> c.d.o.server -> Re: datawarehouse extraction, snapshots?
In my opinion, you have 2 choices.
Think it over and proceed carefully. Your workaround can potentially
get very
expensive and messy.
sg <s4v4g3_at_europe.com> wrote in message news:<3D5C2B32.80607_at_europe.com>...
> the groups who designed the datawarehouse didnt design the processes to
> extract incrementally, yes very sad. That's why I am trying to do it
> with snapshots
>
> The source system is pretty powerful, 3 node TruCluster Server in Compaq
> GS320 but it is not giving enough because as I mentioned same data is
> extracted over and over again everyday plus the billing system itself
> has many batch processes going everynight too!
>
> And yes I am considering replicating the data from source to
> datawarehouse and do all transformations there then distribute data to
> rest of datamarts. I have been doing some research but seems like the
> basic ETL processes are mostly
>
> sqlplus spool
> ftp
> sql loader
>
> that is good if we dont have to deal with quite considerable amount of
> data but since we are not doing anything incrementally everything is
> slow, in order to do extraction delta we probably have to modify source
> data tables (like add a column for last DML date) which seems impossible
> because the billing software provider although they are willing to do it
> but they probably take 4 ~ 8 months! Then we have to test etc which can
> take another couple of months!
>
> I am using 8.1.7 so.... transportable tablespace is not an option
Received on Fri Aug 16 2002 - 01:14:53 CDT