Dept. of Information Engineering, University of Padova, Padova, Italy.
†This work was supported in part by MIUR of Italy under project MAINSTREAM and by the EC/IST Project 15964 AEOLUS. A preliminary version of this work was published in the Proceedings of the Proceedings of the First International Conference on Complex, Intelligent and Software Intensive Systems, Vien, April 2007.
The availability of enormous amounts of un-used computing power and data storage over the In-ternet makes the development of a globally distributed computing platform, called Overlay Computer (OC), at-tractive for the industry, the scientific community, and also the general user. In order to be effective, an OC must be equipped with tools for estimating and optimiz-ing the performance of applications. The tools must rely upon performance metrics of the key factors impacting performance of a distributed application, such as inter-connection latency and bandwidth, and available comput-ing power at the nodes. This paper first reviews the state-of-the-art on performance benchmarking approaches in globally distributed systems, including performance-conscious P2P systems. Later, it addresses the problem of developing a suite of microbenchmarking experiments aimed at providing the basic functionalities of a measurement tool for a P2P-based OC platform. Results show that such a measuring system should take into account the communication patterns generated by the applications in order to provide useful performance insights.
