Analytical Performance Modeling for Computer Systems

This book is an introduction to analytical performance modeling for computer systems, i.e., writing equations to describe their performance behavior. It is accessible to readers who have taken college-level courses in calculus and probability, networking, and operating systems.

This is not a training manual for becoming an expert performance analyst. Rather, the objective is to help the reader construct simple models for analyzing and understanding the systems that they are interested in.

Describing a complicated system abstractly with mathematical equations requires a careful choice of assumptions and approximations. They make the model tractable, but they must not remove essential characteristics of the system, nor introduce spurious properties.

To help the reader understand the choices and their implications, this book discusses the analytical models for 40 research papers. These papers cover a broad range of topics: GPUs and disks, routers and crawling, databases and multimedia, worms and wireless, multicore and cloud, security and energy, etc. An appendix provides many questions for readers to exercise their understanding of the models in these papers.

Y.C. Tay received his BSc from the University of Singapore and PhD from Harvard University. He is a professor in the Department of Mathematics and Department of Computer Science at the National University of Singapore. He has spent sabbaticals at Princeton, MIT, Cambridge, UCLA, Microsoft, Intel and VMware. He has served on program committees for SIGMET-RICS, SIGMOD, PERFORMANCE, VLDB, ICDE, MASCOTS, IWQoS and ICS. His main research interest is performance modeling (database transactions, wireless protocols, trafficequilibrium, cache misses). Other interests include database systems and distributed protocols. He has won several teaching awards.