|
|
|
@inproceedings{DBLP:conf/cikm/Shrufi94,
author = {Adel Shrufi},
title = {Performance of Clustering Policies in Object Bases},
booktitle = {Proceedings of the Third International Conference on Information
and Knowledge Management (CIKM'94), Gaithersburg, Maryland, November
29 - December 2, 1994},
publisher = {ACM},
year = {1994},
pages = {80-87},
ee = {db/conf/cikm/Shrufi94.html, http://doi.acm.org/10.1145/191246.191263},
crossref = {DBLP:conf/cikm/94},
bibsource = {DBLP, http://dblp.uni-trier.de}
}
BibTeX
In this paper, we address the problem of clustering graphs in object-oriented databases. Unlike previous studies which focused only on a workload consisting of a single operation, this study tackles the problem when the workload is a set of operations (method and queries) that occur with a certain probability. Thus, the goal is to minimize the expected cost of an operation in the workload, while maintaining a similarly low cost for each individual operation class.
To this end, we present a new clustering policy based on the nearest-neighbor graph partitioning algorithm. We then demonstrate that this policy provides considerable gains when compared to a suite of well-known clustering policies proposed in the literature. Our results are based on two widely referenced object-oriented database benchmarks; namely, the Tektronix HyperModel and OO7.
Copyright © 1994 by the ACM, Inc., used by permission. Permission to make digital or hard copies is granted provided that copies are not made or distributed for profit or direct commercial advantage, and that copies show this notice on the first page or initial screen of a display along with the full citation.