LISTSERV at the University of Georgia
Menubar Imagemap
Home Browse Manage Request Manuals Register
Previous messageNext messagePrevious in topicNext in topicPrevious by same authorNext by same authorPrevious page (July 2002, week 4)Back to main SAS-L pageJoin or leave SAS-L (or change settings)ReplyPost a new messageSearchProportional fontNon-proportional font
Date:   Wed, 24 Jul 2002 14:16:08 -0400
Reply-To:   Sigurd Hermansen <HERMANS1@WESTAT.COM>
Sender:   "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:   Sigurd Hermansen <HERMANS1@WESTAT.COM>
Subject:   Re: Settle an argument about a view
Comments:   To: Biff Henderson <tango_zebra2002@YAHOO.COM>
Content-Type:   text/plain; charset="iso-8859-1"

A couple of lightning strikes zapped my attempts to reply yesterday. At this stage I have only a couple of thoughts left ....

In my experience a data step (or SQL) view works like an internal pipe. When it executes it streams data from sources and directs them to a display or to another level of a program. It appears to use the same buffers, other memory caches, and temporary files that ordinary SAS data steps and procedures use. On that score, as others have said, Biff rules.

Biff's former friend can still argue effectively against using a view in some situations. Let's say that data updates occur at night, and that during the day an average of five people per day request a summary by region of a large set of data. Rather than read and summarize the data set five times daily, it would make better sense to execute the summary immediately following each overnight update and store the result as a table instead of a view.

Views make better sense when data sources can change continuously. In those cases, creating a work or permanent dataset will serve no particular purpose and will require more resources.

Sig -----Original Message----- From: Biff Henderson [mailto:tango_zebra2002@YAHOO.COM] Sent: Tuesday, July 23, 2002 4:55 PM To: SAS-L@LISTSERV.UGA.EDU Subject: Settle an argument about a view

A colleague of mine insists that when a data step view is called, SAS creates a full physical copy of the "view" somewhere in its workspace prior to the next data step or proc. I say that is not exactly true. I believe that no output data set is created when the view is called. I think each output record is fed into the next data step or proc. (There may be some temporary datasets built, however).

Her arguments against a view was something like "... if 5 people call that view at the same time, SAS has to generate 5 physical copies. Where is that space savings now...?"

Again, I don't believe this to be true. Can some straighten us out?

Thanks, Biff

__________________________________________________ Do You Yahoo!? Yahoo! Health - Feel better, live better http://health.yahoo.com


Back to: Top of message | Previous page | Main SAS-L page