Date:         Thu, 2 Mar 2000 17:48:21 GMT
Reply-To:     "John M. Wildenthal" <jmwildenthal@MY-DEJA.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "John M. Wildenthal" <jmwildenthal@MY-DEJA.COM>
Organization: Deja.com - Before you buy.
Subject:      Re: Performance of Pass Through Facility v/s Views

In article <s8be4b01.058@Fhemaild.fhwa.dot.gov>,
  Yusuf Mohamedshah <Yusuf.Mohamedshah@IGATE.FHWA.DOT.GOV> wrote:
> A number of times I have seen on SAS-L people saying that SQL Pass-
> Through Facility is better than using Views to access the data from a
> RDBMS (such as Oracle or SYBASE). My experience has always been to
> the contrary. Let me illustrate by giving following two examples:
>
> In the first example I am creating a file CA95ROAD with limited
> variables on my PC by accessing the data from Sybase server. The
> first piece of code uses the SQL Pass-Through facility and the second
> piece uses plain Data step with Views.

SAS Version 7 and beyond essentially translates everything it can and
passes it through to the underlying RDBMS.  If it didn't, don't you
think that merge would have taken a bit more CPU if performed on your
machine?  If you think not, copy the two files over into datasets on
your local machine and then merge them.  Your network connection is
almost certainly slower than your hard drives.  The only advantage of
pass-through in v7 is to access RDBMS specific features - usually
performance enhancing features.


Sent via Deja.com http://www.deja.com/
Before you buy.