Date:         Fri, 16 Apr 1999 15:32:19 -0500
Reply-To:     "Lambert, Bob" <Bob_Lambert@AFCC.COM>
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         "Lambert, Bob" <Bob_Lambert@AFCC.COM>
Subject:      FW: Oracle DB and SAS - reply
Content-Type: text/plain

> You might also leverage frequency of data access.  That is, if data most
> often analyzed is that within the last one or two years, put those data in
> sas data sets and the other in views.
>
> I am not an ORACLE or database expert but is there opportunity in the data
> base design?  Are "stagnant" things like Name, Address, SSN, etc being
> stored in tables with dynamic data?  These things might be better stored
> separate table and then the relational features of the database can be
> used to extract them when needed.  Without knowing exactly what your data
> looks like, this is somewhat of a guess but often these simple things
> exist simply because it seemed like a good idea at the start of the
> project.
>
> On this situation where the user would like to have data stored or
> arranged differently (including something as simple as the addition of an
> index), I've been hit with the same reply -- "Sorry, not enough space".
> Again, I don't know your specific situation, but sometimes I feel like the
> DBA isn't tied into the big picture.  The database they're managing is for
> the purpose of extracting information which is to be used in decision
> making, which in turn affects the bottom line in some way -- revenue
> enhancement, expense reduction, etc.  This is likely no fault of their
> own.  In short, help them see the big picture if necessary.  They will be
> more likely to re-assess the "not enough space" response from a different
> vantage point.  (Don't bet the farm on it, though).
>
> Bob Lambert
> Associates Capital Bank.
>
> -----Original Message-----
> From: Anne Marie Smith [SMTP:rainbowsoftware@PRODIGY.NET]
> Sent: Friday, April 16, 1999 8:54 AM
> Subject:      Oracle DB and SAS - reply
>
> Hi Jules,
>    This is a good suggestion because seperate views created with SAS's
> Proc SQl  for each of the 5 years would not take up much space.  But I
> think one really needs to know how much longitudinal data is involved as
> to whether it would be better to have all five years in one table or
> seperate tables for each year.  If its massive amounts then by all means
> seperate it by year or even 6 month intervals.  While doing report
> writing, I've found that one can only get about 6 months worth of counts,
> percents and some stats on longitudinal data on one page.
>
> HTH,
> Anne Marie Smith at  Rainbow Communications and Software Inc.
> <http://www.rainbow-software.com>
>
>
> --
> ned!
>   << File: Card for Anne Marie Smith >>