Re: Oracle (or Informix, DB2) vs. SAS for Enterprise Warehouse

I am an Oracle fan, but there are some things SAS is good for - number crunching being the obvious one. I have worked with SAS users who think because SAS is good for their purposes it must be a good all round tool or even database server. It does not have a good security system, SAS does not offer the same support for database action (row locking, rollbacks, etc), less tools could talk to SAS than Oracle (e.g. OLAP, ODBC),

In my last position they pushed me to learn SAS (sent me on training, etc) but I could not get it to do anything useful because _I_ lacked the mathematical abilty and knowledge to use its number crunching abilty well.

We did quite a bit of selecting out of an Oracle warehouse into SAS datasets for analysis. This made sense to me. But then we built a historical warehouse in SAS. The organisation is now redesigning the original Oracle data warehouse to hold historical data.

My personal view is hold the data in a database and extract it into semi-temporay datasets for analysis. If the SAS script is retained and the data is refreshed from a consistent database the analysis is replicatable. This offers the best of both - the robustness and multi-user capability of the database and the mathematical functionality of SAS without thrashing the database server.

On the other hand there are considerable overheads with, in particular, Oracle (rollback segments, pctfree, pctused, etc) which means a database solution is likely to require more disk space than a SAS solution.

my 10cents worth
Karl

Jack Sanderson <jogault_at_hotmail.com> wrote in message news:8df57e$lc2$1_at_nw001t.infi.net...
> There's a debate raging here on the better very scalable data warehouse
> environment.
> The SAS camp claims that SAS is better for building predictive models,
> easier to work w/external data sets, robust enough for high volume/high
> scale.
> The RDB camp (to which I belong) says that the RDB offers higher
> scalability, better utilization of parallel processors, and much better
> openess to OLAP and CRM tools.
>
> Thoughts?
>
>
>
>
Received on Tue Apr 18 2000 - 00:00:00 CEST