Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!spool.mu.edu!agate!eden.Berkeley.EDU!mao
From: mao@eden.Berkeley.EDU (Mike Olson)
Newsgroups: comp.databases
Subject: Re: Rushmore Query Optimization in FoxPro 2.0
Message-ID: <1991Jun27.225122.26594@agate.berkeley.edu>
Date: 27 Jun 91 22:51:22 GMT
References: <91178.110048RMC100@psuvm.psu.edu>
Sender: usenet@agate.berkeley.edu (USENET Administrator)
Organization: University of California at Berkeley
Lines: 37

In <91178.110048RMC100@psuvm.psu.edu>, RMC100@psuvm.psu.edu (Randy Carraghan)
writes

> I'm not a FoxPro user, but I was interested in finding out more about the
> Rushmore Query Optimization they plan to use in release 2.0 of their product.
> The advertisement I read said that a patent had been applied for [ ... ]
> Are these algorithms
> available in publication?  I'd like to see some of the techniques used to
> fulfill the SQL queries.

i asked a similar question about a month ago.  a couple of people responded
to me.  the concensus was that rushmore uses at least two techniques to
speed up query processing.  one strategy is to index only certain ranges of
values for a given attribute in a relation.  for certain query patterns,
indexing ten percent of your table could speed up ninety percent of your
queries.  the other strategy was to optimize data dictionary operations
inside the planner and optimizer in the same way as normal queries.  many
systems hard-code table accesses internal to the dbms engine.  if you optimize
the table accesses done by a 'define index' command, you may be able to make
it go faster.

prior art exists for both of these techniques, so if this is correct, then
there's nothing especially interesting about rushmore.  in fact, i would
wager that the patent application is more for the benefit of the marketing
department than to protect the developments of the engineering department.

i should emphasize that i have no connection with foxpro, and that none of
my correspondents was willing to state that his or her information was
authoritative.  this may or may not be an accurate description of some part
of rushmore.  given that a patent application is outstanding, however, this
is probably the most authoritative answer you'll get for a while.  if anyone
with solid information can share it, publically or privately, i would be
happy to see it.
					mike olson
					postgres research group
					uc berkeley
					mao@postgres.berkeley.edu