Modelling overhead of tuple spaces with design of experiments

Authors:
Frederic Hancke;Tom Dhaene;Jan Broeckhove
Affiliations:
University of Antwerp, Antwerp, Belgium;University of Antwerp, Antwerp, Belgium;University of Antwerp, Antwerp, Belgium
Venue:
VECPAR'04 Proceedings of the 6th international conference on High Performance Computing for Computational Science
Year:
2004

Citing 6
Cited 0

Generative communication in Linda

ACM Transactions on Programming Languages and Systems (TOPLAS)
Performance prediction and tuning of parallel programs

Performance prediction and tuning of parallel programs
Performance prediction and scheduling for parallel applications on multi-user clusters

Performance prediction and scheduling for parallel applications on multi-user clusters
JavaSpaces Principles, Patterns, and Practice

JavaSpaces Principles, Patterns, and Practice
Core JINI

Core JINI
Design and Analysis of Experiments

Design and Analysis of Experiments

Quantified Score

Hi-index	0.00

Visualization

Abstract

In the world of high performance distributed computing new, better and faster distributed platforms evolve quite rapidly. Building application dependent performance test cases in order to make platform comparisons, is not satisfying anymore. This paper presents the use of a statistical technique, called Design of Experiments (DoE), to model the performance of two closely related distributed platforms, JavaSpaces and GigaSpaces, with as few tests as possible. All tests will be based on Tunable Abstract Problem Profiles (TAPPs), a technique for problem independent simulations.