[ont.events] UW Data Structuring Seminar, Dr. Yu on "Distributed Query Processing"

mwang@watmath.UUCP (mwang) (06/19/85)

DDEEPPAARRTTMMEENNTT OOFF CCOOMMPPUUTTEERR SSCCIIEENNCCEE
UUNNIIVVEERRSSIITTYY OOFF WWAATTEERRLLOOOO
SSEEMMIINNAARR AACCTTIIVVIITTIIEESS

DDAATTAA SSTTRRUUCCTTUURRIINNGG SSEEMMIINNAARR

                    - Monday, June 24, 1985.

Dr. C. Yu of the University of Illinois at Chicago will
speak on ``Distributed Query Processing.''

TTIIMMEE::                3:30 PM

RROOOOMM::              MC 5158

AABBSSTTRRAACCTT

In  this  talk, two different algorithms are presented.
One  algorithm  makes  use  of semi-joins to reduce the
amount  of  data transfer.  Other features of the algo-
rithms are

*    dynamic execution of semi-joins

*    elimination of useless relations

*    replacement of some semi-joins by better ones

*    making  use  of  redundant  copies of relations to
     reduce communication cost.

     The  other algorithm is a ``fragment and replicate
algorithm''.  The key features of the algorithm are the
use  of semantic information to process queries locally
without  data  transfer  and  taking into consideration
that the presence or absence of fast access paths (e.g.
indices)  can  be  a  major factor in determining local
processing cost.

     The  project  is a joint research project with the
System  Development Corporation.  Future development of
the project will be discussed.

                          June 19, 1985

mwang@watmath.UUCP (mwang) (06/20/85)

DEPARTMENT OF COMPUTER SCIENCE
UNIVERSITY OF WATERLOO
SEMINAR ACTIVITIES

DATA STRUCTURING SEMINAR

                    - Monday, June 24, 1985.

Dr. C. Yu of the University of Illinois at Chicago will
speak on ``Distributed Query Processing.''

TIME:                3:30 PM

ROOM:              MC 5158

ABSTRACT

In  this  talk, two different algorithms are presented.
One  algorithm  makes  use  of semi-joins to reduce the
amount  of  data transfer.  Other features of the algo-
rithms are

*    dynamic execution of semi-joins

*    elimination of useless relations

*    replacement of some semi-joins by better ones

*    making  use  of  redundant  copies of relations to
     reduce communication cost.

     The  other algorithm is a ``fragment and replicate
algorithm''.  The key features of the algorithm are the
use  of semantic information to process queries locally
without  data  transfer  and  taking into consideration
that the presence or absence of fast access paths (e.g.
indices)  can  be  a  major factor in determining local
processing cost.

     The  project  is a joint research project with the
System  Development Corporation.  Future development of
the project will be discussed.

                          June 20, 1985