1 © 2007 V.A.F. Almeida and D.A. Menascé. All Rights Reserved. 1 Web Workloads: characterization, modeling, and application Virgilio A.F. Almeida Federal University of Minas Gerais www.dcc.ufmg.br/~virgilio and Daniel A. Menasce George Mason University www.cs.gmu.edu/faculty/menasce.html WWW 2007 Banff, Canada © 2007 V.A.F. Almeida and D.A. Menascé. All Rights Reserved. 2 Copyright Notice • The copyright to this set of slides belongs to Virgilio A.F. Almeida and Daniel A. Menasce.These slides were prepared for the participants of a tutorial presented at the WWW 2007 Conference. Copying, distributing, posting on a Web site, in part or in its entirety, requires written permission of the authors. © 2007 V.A.F. Almeida and D.A. Menascé. All Rights Reserved. 3 What is a workload? A bridge designer needs to know: -the bridge’s type of service (e.g., highway, railroad, pedestrian) -the bridge’s construction material (e.g., steel) -the bridge’s operating rating, i.e., the absolute maximum permissible load level per vehicle (e.g., 44.1 metric tons). - average daily traffic (e.g., 5,400 cars, 1,200 trucks) The bridge designer needs to understand and quantify the bridge’s workload: type of load and load levels. © 2007 V.A.F. Almeida and D.A. Menascé. All Rights Reserved. 4 What is a workload? A Web site’s designer needs to know: -The web site’s type of service (e.g.,public, private, commerce) -The site’s hardware and software platform (e.g., Linux on Intel servers) -the site’s operating rating, i.e., the absolute maximum of concurrent connections (e.g., 300 concurrent connections) -average daily traffic (e.g., 2million HTTP requests, 30K SSL connections) The site’s designer needs to understand and quantify the site’s workload: type of load and load levels. © 2007 V.A.F. Almeida and D.A. Menascé. All Rights Reserved. 5 Outline 1. Introduction 2. Basic concepts 3. Advanced concepts 4. E-Business workloads 5. Blog workloads 6. Search engine workloads 7. Auction workloads 8. Spam workloads 9. Streaming media workloads 10. Applications to design 11. Conclusions © 2007 V.A.F. Almeida and D.A. Menascé. All Rights Reserved. 6 Workload • The workload of a system is the set of all inputs the system receives from its environment during any given period of time. HTTP requests Web Server