Date:         Sun, 23 Feb 2003 07:58:02 -0600
Reply-To:     GSnell@DataSavantConsulting.com
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Gregg P. Snell" <GSnell@DATASAVANTCONSULTING.COM>
Subject:      Re: multi thread sas in Datawarehousing environment
Comments: To: scott sanders <scsand@ACXIOM.COM>
In-Reply-To:  <a5eg5vsqi0slfrj4sb3aduj3hrsfrb9bk1@4ax.com>
Content-Type: text/plain; charset="us-ascii"

Scott,

If your host has CONNECT licensed (which it probably does) then using MP
CONNECT is what you want. Some great examples are here:
http://support.sas.com/rnd/scalability/tricks/index.html

Regards,

Gregg P. Snell
(913) 638-4640
(208) 977-1943 efax
http://www.DataSavantConsulting.com

-----Original Message-----
From: SAS(r) Discussion [mailto:SAS-L@LISTSERV.UGA.EDU]On Behalf Of
scott sanders
Sent: Saturday, February 22, 2003 9:10 PM
To: SAS-L@LISTSERV.UGA.EDU
Subject: multi thread sas in Datawarehousing environment


I work in a datawarehouseing environment where I have to execute sas
code against hundreds of millions of rows.

The process I have now works well

1. Create a table in oracle with all my attributes necessary for sas
excecution
2. Create another table in oracle with 20 partitions (essentially same
# of obs per partition)
3.  Kick of 1 sas job per partition (20).

It has the effect of multi-threading and runs fast.

My question:

If I could only create one table without any partitions....

1.  How could SAS look at a unique key and split the data based on
that key.
2.  Can one sas job spawn numerous other's to accomplish what I am
doing executing sas in a unix env. (#3 above).

Any ideas would be greatly appreciated.  Thanks in advance.

scott