Performance Guarantees for Web Server End-Systems: A Control-Theoretical Approach Tarek F. Abdelzaher Department of Computer Science University of Virginia Charlottesville, VA 22903 Kang G. Shin Real-Time Computing Laboratory EECS Department, University of Michigan Ann Arbor, MI 48109–2122 Nina Bhatti Hewlett Packard Laboratories 1501 Page Mill Road Palo Alto, CA 94304 Abstract The Internet is undergoing substantial changes from a communication and browsing infrastructure to a medium for conducting business and marketing a myriad of services. The World Wide Web provides a uniform and widely-accepted application interface used by these ser- vices to reach multitudes of clients. These changes place the web server at the center of a gradually emerging e- service infrastructure with increasing requirements for service quality and reliability guarantees in an unpre- dictable and highly-dynamic environment. This paper describes performance control of a web server using classical feedback control theory. We use feedback control theory to achieve overload protection, performance guarantees, and service differentiation in the presence of load unpredictability. We show that feed- back control theory offers a promising analytic foun- dation for providing service differentiation and perfor- mance guarantees. We demonstrate how a general web server may be modeled for purposes of performance control, present the equivalents of sensors and actuators, formulate a simple feedback loop, describe how it can leverage on real-time scheduling and feedback-control theories to achieve per-class response-time and throughput guar- antees, and evaluate the efficacy of the scheme on an experimental testbed using the most popular web server, The work reported in this paper was supported in part by the NSF under Grant EIA-9806280. Apache. Experimental results indicate that control- theoretic techniques offer a sound way of achieving de- sired performance in performance-critical Internet ap- plications. Our QoS (Quality-of-Service) management solutions can be implemented either in middleware that is transparent to the server, or as a library called by server code. Keywords: Quality of Service, Web Servers, Control Theory, Performance Guarantees 1 Introduction The Internet has become an important medium for con- ducting business and selling & buying services. The web presents a convenient interface for the emerging performance-critical applications such as online trading and e-commerce. These applications require stringent performance guarantees from the web server (e.g., that an online trade will be executed in a timely manner to avoid potential financial loss). Attainment of these guar- antees is especially difficult due to the unpredictable na- ture of the Internet and server load. In this paper, we show how feedback control theory can be used as an ana- lytic engine to provide robust performance guarantees in the presence of load and resource uncertainty. Feedback control theory was originally developed for physical process control. Its use in the context of software per- formance control is novel. Software performance con- trol presents a myriad of interesting challenges such as selecting proper software sensors and actuators, model- 1