Performance evaluation of multi-tier Web clusters
The need to optimize the performance of Web-based services is producing a variety of novel architectures for high performance and scalable servers and networks, ranging from Web clusters distributed in a LAN environment, to multi-clusters distributed over the Internet, to Content Distribution Networks.
One of the key issues is the evaluation of the performance and scalability of these systems under realistic workload conditions. The following research proposal identifies the main requirements associated with a benchmarking study of distributed architectures supporting highly accessed Web sites.
We discuss the performance tradeoffs of various options in the design and operation of scalable Web systems through the presentation of some case studies related to locally distributed systems, including the performance analysis of multi-tier architectures.
Our study leads us to conclude that most of the considered benchmarking tools work properly when they are used to analyze a single server system (in many cases even a single-tier architecture), but none of them is able to address all issues related to the performance analysis of the most recent distributed and multi-tier Web systems.
This page includes a selected list of publications since 2002. Publications not available online can be requested to the authors.