![]() |
![]() |
This tutorial has been CANCELLED. M3 Deploying and Benchmarking Web Caches Peter Danzig, Network Appliance and USC and Alex Rousskov, National Laboratory for Applied Network Research (NLANR) Who should attend: Engineers, managers, developers and system administrators deploying web cache solutions for ISPs, carriers, and large enterprises. Participants should know the basics of web caching, web protocols and be interested in deploying and evaluating web caches for 100Mbps, DS3, and higher bandwidths. The first half of this tutorial covers the issues of sizing individual cache nodes, and then how to aggregate cache nodes to scale to hundreds of megabits of WAN bandwidth. We will describe L4 switches for partitioning client traffic and clustering individual cache nodes, and also will describe how to build a distributed web server from proximity-detecting load balancers and web caches. We will also review current statistics from operating web caches. We will describe new applications of web caches to network news (NNTP) distribution and streaming media distribution.
The second half of this tutorial complements the first by covering a wide range of benchmarking activities including:
Several state-of-the-art benchmarks will be discussed and compared. Both white box and black box testing techniques will be studied. The presentation is based on numerous real world test cases and sample benchmarking sessions. After completing this tutorial, participants will understand what is involved in building a robust, high volume web cache, and will be able to benchmark it and be confident that their results are solid and conclusions are valid. The audience will receive essential information about interpreting benchmark results and will leave with a ready-to-use collection of benchmarking tricks, caveats, and pitfalls. ![]()
|
| |||||
|
![]() ![]() ![]() |