Date: Wed, 21 Nov 2001 12:22:14 -0500
Reply-To: "Jeffrey P. Crabb" <jeff_crabb@MASTERCARD.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: "Jeffrey P. Crabb" <jeff_crabb@MASTERCARD.COM>
Subject: Re: Sas vs. MSSQL
On Tue, 20 Nov 2001 11:55:06 -0800, Miroslav <smico@HOTMAIL.COM> wrote:
>We have been considering purchase of relation database to move few
>databases (largest is about 70MM records - 1.4GB binary file).
>
>As we do lots of statistical research, SAS has become one of the
>candidate. Other candidate is MS SQL 2000 Server (25 CAL).
>
>I was trying to search SAS' WEB site, but couldn't get some answers.
>While I am waiting for their sales reps. to call me back, I would like
>to ask SAS gurus a few questions:
>
>1. Can I set-up replication of SAS database between remote servers?
>How reliable it is?
>2. How much SAS charge in first, and consecutive years?
>3. Is there still limit of 2GB on database size for NTFS for certain
>SAS procedures
>( - as per their site:
>http://www.sas.com/partners/directory/intel/faq.html:
>
>What are the largest data sets the SAS System can process on an NT
>Workstation or Server?
>
>Using the NTFS (NT File System) file system, the SAS System on NT can
>process 16 petabytes of data in the data step, and the contents,
>cimport, cport, freq, sort, and summary procedures. For all other
>procedures, using the NTFS file system, the SAS System on NT can
>process 2 gigabytes of data. Using the FAT (File Allocation Table)
>file system, the SAS System on NT can process 2 gigabytes of data
> )
>
>4. I would like to create SAS OLE objects and control SAS from
>Fortran, C++ and Delphi programs. How deep their OLE interface covers
>SAS functionality? I know that I can run SQL Querys...
>
>Thank in advance for your help
>Miroslav
Miroslav,
I can answer a part of your questions. If you're going to use SAS as a
relational database product, you may not be happy with your results.
SAS is an analysis tool, not a relational database tool. Decide what you
want or need. If you want/need a relational database tool then use Oracle
or one of its derivatives (I'm not a big fan of Microsoft for much of
anything). If you want/need an analysis tool, use SAS.
SAS should be able to handle your files, provided you have enough SAS work
disk space and machine memory. Most of my time is spent with UNIX for
heavy SAS number crunching and NT for simple SAS analysis and graphing.
With 70MM records, don't you think you need to migrate to UNIX? I
routinely handle files over 2GB on UNIX and often go to the 40GB range. I
wouldn't want to do any of that processing on an NT machine -- way too slow!
Jeff Crabb
Statistical Services
MCIdeas
MasterCard International
|