|
|
|
@inproceedings{DBLP:conf/dasfaa/KitsuregawaNT89,
author = {Masaru Kitsuregawa and
Miyuki Nakano and
Mikio Takagi},
editor = {Sukho Lee and
Hideko S. Kunii and
Won Kim and
In Sup Paik and
Yahiko Kambayashi},
title = {Funtional Disk System as a High Performance Relational Storage},
booktitle = {International Symposium on Database Systems for Advanced Applications,
Seoul, Korea, April 10-12, 1989},
publisher = {Dept. of Computer Science, KAIST, P.O. Box 150, ChongRyang, Seoul,
131-650, Korea},
year = {1989},
pages = {243-250},
ee = {db/conf/dasfaa/KitsuregawaNT89.html},
crossref = {DBLP:conf/dasfaa/89},
bibsource = {DBLP, http://dblp.uni-trier.de}
}
Functional Disk System with Relational database engine(FDS-R) is a relational storage system designed to accelerate relational algebraic operations. FDS-R employs a filtering and dynamic clustering mechanism as a special hardware function and provides an intelligent data management and an efficient data processing. The relation size which could be handled on the first prototype, however, was limited to the size of a staging buffer. Then we built up the second version of FDS-R, FDS-RII, which is designed to handle large relations efficiently. We have presented the processing strategy: the "Extended Task Cycle" for relational algebraic operations on FDS-RII, where the strategy is selected at run time form two algorithms (Nested Loop and Grace Hash Algorithms) by comparing the estimated I/O costs of them.
In this paper, we present the overview of Hardware configuration and Software system of FDS-RII. First we show the basic Task Cycle and give the performance evaluation results using the original Wisconsin Benchmark for small relations, where filtered relation can be staged on the staging buffer. FDS attains higher performance than the current database systems such as INGRES, Oracle and IDM. Next we explain the "Extended Task Cycle" which was introduce to handle the relations larger than the staging buffer size. With the expanded version of Wisconsin Benchmark, we measured the FDS-RI1 performance. FDS-RII attained a high performance for processing large relations as compared to other large database systems such as Gamma and Teradata. While FDS-RII uses just one disk and three MC68020, Teradata uses 40 disks and 20 AMP's and Gamma requires 8 disks and 17 VAX 11/75O's.
Copyright © 1989 by The Organizing Commitee of the International Symposium on Database Systems for Advanced Applications. Permission to copy without all or part of this material is granted provided that the copies are not made or distributed for direct commercial advantage, the DASFAA copyright notice and the title of the publication and its date appear, and notice is given that copying is by permission of the Organizing Commitee of the International Symposium on Database Systems for Advanced Applications. To copy otherwise, or to republish, requires a fee and/or special permission from the Organizing Commitee.
