Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site water.UUCP
Path: utzoo!watmath!watnot!water!ylfink
From: ylfink@water.UUCP (ylfink)
Newsgroups: ont.events
Subject: UW CS Colloq. Semi., Dr. Sellis on "Optimization of Extended Relational Database Systems".
Message-ID: <208@water.UUCP>
Date: Tue, 11-Feb-86 13:30:49 EST
Article-I.D.: water.208
Posted: Tue Feb 11 13:30:49 1986
Date-Received: Wed, 12-Feb-86 20:22:00 EST
Expires: Tue, 18-Feb-86 00:00:00 EST
Organization: U of Waterloo, Ontario
Lines: 44


DEPARTMENT OF COMPUTER SCIENCE
UNIVERSITY OF WATERLOO
SEMINAR ACTIVITIES

COMPUTER SCIENCE COLLOQUIUM

                    - Monday, February 17, 1986.

Dr.  Timos  Sellis  of  the  University  of California,
Berkeley will speak on ``Optimization of Extended Rela-
tional Database Systems''.

TIME:                3:30 PM

ROOM:              MC 1050

ABSTRACT

In  this  talk  I  will  examine  some query processing
optimization  problems  that arise in an extended rela-
tional  database  management  system.   These  extended
relational  systems aim at increasing the functionality
of  the  existing systems in support of engineering and
AI  applications.  First, I will describe the prototype
environment,  QUEL+ (extended QUEL), which motivates my
research.   The  two  major enhancements of QUEL+ are *
storing  relational expressions in relation fields, and
*  repetitive  execution  of commands. The implications
that these extensions have on query processing are dis-
cussed and some conventional query optimization tactics
are  proposed for this extended language.  Then, I will
elaborate  on the problem of optimizing a collection of
commands  rather than one command at a time.  Such mul-
tiple command processing arises in the above context if
a  field  contains more than one query language expres-
sion  that  has  to be evaluated.  Optimization tactics
based  on  reordering  the  sequence of commands and on
transforming  the  sequence into a faster-executing but
semantically  equivalent  sequence  of commands will be
presented.   Finally,  for  processing  collections  of
retrieve-only  commands,  I will describe some multiple
query  optimization algorithms that aim at reducing the
number of disk accesses.