Date:         Tue, 9 Jun 1998 16:16:36 -0500
Reply-To:     Rob Rohrbough <rob@ROHRBOUGH-SYSTEMS.COM>
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         Rob Rohrbough <rob@ROHRBOUGH-SYSTEMS.COM>
Organization: Rohrbough Systems Design, Inc.
Subject:      Re: SAS-L Digest - 9 Jun 1998 - Special issue
Content-Type: multipart/mixed;

James,
 
1-25M is a pretty wide range.  I currently am working with a 1.5M obs by
100 variable application on a 266Mhz Pentium with 14GB of RAID-5 (not
the fastest choice) and two Ultra-IDE spindles one of which is 6.4GB and
holds SASWORK.  While a PPro or one of the new 400Mhz Intels with faster
bus and internal cache may help, our secret has been in the HD
management and SAS programming techniques for efficiency.  We are also
trying to work with an IDE RAID-0.
 
BTW, there are 6 more variables related to this master in a second file
of about 2.5M obs.  The join can cause real performance problems.  We
have tricked PROC SQL into using a merge and are trying to get to a
merge datastep.  For HD managment, it seems the more spindles, the
better.  For sure, separate your work area from your master files.
 
HTH,
 
Rob
 
 
------------------------------------------------------------------------
James Yang wrote:
>
> Is it possible to process and store 1-25 milions of observations (10-30
> variables) on a powerful PC/workstation using SAS. If so, what hardwares
> are required (1 or more than 1 tape drives)?
>
> Any response will be greatly appreciated!
>
> James
[text/x-vcard]

begin:          vcard
fn:             Rob Rohrbough
n:              Rohrbough;Rob
org:            Rohrbough Systems Design, Inc.
adr:            9215 Dorcas Street;;;Omaha;NE;68124-2039;USA
email;internet: rob@rohrbough-systems.com
title:          Consultant
tel;work:       (402) 343-1493
tel;home:       (402) 397-6915
note:           A SAS Institute Quality Partner
x-mozilla-cpt:  ;0
x-mozilla-html: FALSE
version:        2.1
end:            vcard