Performance guarantees for Web server end-systems: a control-theoretical approach

Abstract

AbstractÐThe 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 services 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 unpredictable 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 feedback control theory offers a promising analytic foundation for providing service differentiation and performance 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 guarantees, and evaluate the efficacy of the scheme on an experimental testbed using the most popular Web server, Apache. Experimental results indicate that controltheoretic techniques offer a sound way of achieving desired performance in performance-critical Internet applications. 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.

A Cluster-Based Web System Providing Differentiated and Guaranteed Services

“…1 The main differences concern the introduction of the HTTP/1.1 protocol, and various modifications on clients behavior because the original design of Webstone was not intended to produce realistic workload, but the heaviest load on the server nodes. Basically, the modifications are inspired to the SURGE model [6,15], that is integrated also with user classes and performance metrics oriented to QoS.…”

“…To enforce SLA constraints, Pandey et al [24] examine selective allocation of server resources by assigning different priorities to the page requests. In [1] a control-based approach is proposed for Web service differentiation.…”

Control Theoretic Modelling and Design of Admission Control Mechanisms for Server Systems

“…In [20] a queue length control was developed for an M/M/1-system using stochastic control theory. In [1] and [2] a web server was modelled as a static gain to find controller parameters for a PI-controller. A scheduling algorithm for an Apache web server was designed using system identification methods and linear control theory in [17].…”

Decay function model for resource configuration and adaptive allocation on internet servers

“…On another track altogether, Abdelzaher, et al recently developed feedback control approaches to deal with the impact of high order moments of Internet traffic [1]. They treated requests as aperiodic real-time tasks with arbitrary arrival times, computation times, and relative deadlines.…”

“…However, they are insufficient for the study of the impact of general Internet traffic, particularly its inherent bursty traffic patterns and auto-correlations on the server performance. The following practical yet challenging problems, are largely remaining unsolved, or cannot be precisely resolved in theory: (1) What resource capacity should a server be configured for a requirement of certain levels of QoS? (2) What level of QoS can a server with certain resource capacity support?…”