Who should attend: Anyone responsible for running a Web site or intranet who wants to understand how to improve its performance.

Popular Web sites receive millions of hits per day, but this popularity comes at the expense of network and server delays. Speed and performance are the top problems facing Web users. What issues affect the performance of Web servers and intranets? How do you analyze the components and protocols involved in the execution of Web services?

Once you understand the performance issues, you will learn how to characterize the workload for intranets and Web servers, taking into consideration new phenomena such as "burstiness" and "heavy-tailed probability distributions". You will learn how to develop and use performance models based on queuing networks to identify and analyze problems. Finally, we will discuss some simple examples using queuing network models to identify bottlenecks and pinpoint performance problems in Web environments.

Topics will include:

	Web and intranet performance issues    Where are the delays?    Perception of performance    Infrastructure (basic components, cache, proxy, and mirror)    Web server architecture and workload    Internet and intranet (bandwidth, latency, and traffic)
	Understanding and characterizing the workload    Workload for an intranet    Workload characterization methodology    Heavy-tailed distributions    Bursty workload
	Web performance modeling    Fundamentals of queuing models    System level models    Component level models: open and closed systems    An intranet model    A web server model    A network model

 

Virgilio F. Almeida  is a professor of computer science at Federal University of Minas Gerais, Brazil. He has published over 45 papers in refereed journals and conference proceedings and four books. He received several awards including best paper in the Compaq Technology Award in 1996. He was a visiting professor at Boston University during 1996 and a Visiting Researcher at XEROX PARC in 1997.