Held in conjunction with SIGMETRICS/Performance 2010
Monday, June 14, 2010 - Columbia University, New York
Sponsored by ACM SIGMETRICS
8:30-8:45 Opening Remarks and Introduction
8:45-9:30 Invited talk: Green Data Center Program
Alan Crosswell, Associate Vice President and Chief Technologist for Columbia University Information Technology (CUIT)
9:30-10:10 Session 1: Measurement-based Modeling and Analysis
10:10-10:30 Coffee break
10:30-11:10 Session 2: Wireless
11:10-11:50 Session 3: Networks
11:50-12:30 Lunch
12:30-13:25 Keynote talk: "On Efficient Delivery of Web Content", Mangesh Kasbekar, Principal architect with the service performance department of Akamai Technologies,
13:25-13:30 Closing comments
The Akamai Content Delivery Network is a massive distributed system with tens of thousands of webservers located in nearly every important city and network in the world. These machines serve petabytes of traffic of today's most popular websites on a daily basis. The energy consumed by these servers forms the largest fraction of Akamai's total energy consumption. In this talk, I will focus on the various large trends that have emerged in the Internet traffic over the years, the different ways in which they use the Akamai platform, and the different levels of server efficiency involved in delivering these classes of traffic. With respect to these classes, we will explore the opportunities in improving the energy efficiency of delivering Internet traffic, as well as the challenges and tradeoffs involved in doing so.
Biography: Mangesh Kasbekar works for Akamai Technologies and is a principal architect with the service performance department. His main focus over the years has been the end-to-end performance and availability of the Akamai platform. He also works extensively on improving the efficiency of content delivery for the largest customers and architecture of custom projects. Prior to joining Akamai, he earned a Ph.D. in computer science from the Pennsylvania State University.
In 2008, Columbia University competed for and was awarded a contract from the New York State Energy Research and Development Authority (NYSERDA) to:
The $1.2M project began in April 2009 and will conclude in April 2011 and is attempting to:
This talk will highlight project progress to date, including identifying low-hanging fruit that others can pursue in their data centers as well as some of the surprises one finds when digging into the innards of a 1960's data center.
Biography: Alan Crosswell is Associate Vice President and Chief Technologist for Columbia University Information Technology (CUIT). In this role, he is responsible for a staff of approximately 100 professionals providing telephony, network, desktop, systems infrastructure and research computing and administration services for approximately 40,000 end users. He is a graduate of Columbia University's Fu School of Engineering and Applied Sciences department of computer science (B.S. 1981, M.S. 1984) and has worked at the same Columbia department since being hired as a student computer consultant in the late 1970's. His mother no longer asks when he will get a real job. Mr. Crosswell is a member of the Ivy Plus Networks and Infrastructure groups, the Common Solutions Group, the Internet2 Community Leaders Forum as well as the VoIP, IPv6, Presence and Integrated Communications, and Multicast Working Groups (former chairperson), the EDUCAUSE Campus Cyberinfrastructure Working Group, and Columbia University's Environmental Sustainability Advisory Committee. He has served on several ad-hoc policy and technical committees of the New York State Education and Research Network (NYSERNet), including involvement in development of the NYC dark fiber network, selection of ISPs, and the Syracuse Data Center. Mr. Crosswell is currently Principal Investigator for a $1.2M project co-funded by Columbia University and the New York State Energy Research and Development Authority (NYSERDA) for a green data center pilot project that focuses on development and application of measurement and verification of energy efficiency improvements in the context of an active university data center.