COMPUTER SCIENCE COLLOQUIUM
Characterizing YouTube Workload at a Campus Network
Anirban Mahanti
Department of Computer Science and Engineering
Indian Institute of Technology (IIT), Delhi, India
Thursday, 23 October, 2008 at 10:15
Auditorium U49E
ABSTRACT
This talk discusses the traffic characteristics of the
popular Web 2.0 video sharing service, YouTube. Over a three
month period, we observed over 23 million transactions between
users on a large university network and YouTube, including more
than 600,000 video downloads. At the same time, we collected
statistics on the globally popular videos on YouTube.
In this talk, we examine usage patterns, file properties,
file popularity and referencing characteristics, and
transfer behavior of YouTube. We also study several YouTube user session
characteristics such as session durations and inter-transaction
times. Where appropriate, we will discuss the similarities/differences
between YouTube workload and traditional Web and media streaming
workloads. Implications of our findings on network management, content
distribution strategies, and design of next-generation synthetic
Web workloads are considered. For example, we show that as with
traditional Web, caching could improve the end user experience, reduce
network bandwidth consumption, and reduce the load on YouTube's
infrastructure. Unlike traditional Web caching, however, Web 2.0
provides additional meta-data that should be exploited to improve the
effectiveness of strategies like caching. We conclude the talk with a
brief discussion of current and future work.
While most of this talk will focus on YouTube workload characterization,
a brief overview of my research interests and recent, significant,
research contributions will also be presented.
SDU HOME |
IMADA HOME |
Previous Page
Daniel Merkle