Abstract |
Web 2.0 reflects the current practices in web page design and usage that turned the web into
an application platform and enabled users to create social communities. In this dissertation
we examine two widely used mechanisms that enable file and information sharing in web 2.0.
Furthermore, we illustrate the design and applicability of a new technique for performing
distributed measurement and monitoring tasks.
First, we look at One-Click Hosting (OCH) services, a relatively new type of services,
such as RapidShare and MegaUpload, which offer users the ability to share files through cen-
tralized servers. OCH services are continuously increasing and challenge the dominant usage
of peer-to-peer (p2p) systems for file sharing. Our study uses a combination of passive and
active measurements, and tries to understand OCH service’s architecture, usage patterns and
content characteristics. Furthermore, we compare RapidShare, the leading OCH platform,
with BitTorrent in terms of user-perceived throughput and content availability, and explore
the characteristics of some popular RapidShare indexing sites, a fundamental component for
content discovery.
Second, we provide a wide characterization study of URL shortening services, a number of
services, especially popular within online social networks, that keep a mapping of submitted
URLs to shorter equivalents and redirect users to the original URL upon request. Using
traces of short URLs collected both from crawling a number of URL shortening services, and
collecting short URLs that appear in Twitter messages, we analyze the usage of short URLs
in general and in the context of certain social communities. Furthermore, we challenge the
performance benefits and implications of their wide usage.
Finally, we present MOR, a technique for performing distributed measurement and mon-
iii
itoring tasks using the geographically diverse infrastructure of the Tor anonymizing network.
MOR evolved during our study of One-Click Hosting services and the need for examining
their infrastructure properties. Not limited to that analysis, we also provide a number of
case studies that show the applicability and value of MOR in examining Internet services and
detecting network neutrality violations.
The results of this research are the following. First, we show that OCH services have all
the means to compete with p2p systems for becoming the dominant file sharing platform.
While built on a typical web data-center infrastructure, their usage patterns suggest little
benefit from deploying a more distributed infrastructure. Second, through the study of URL
shortening services we observe that short URLs reflect an “alternative” web, with different
popularity and lifetime patterns, compared to access patterns as seen for the traditional web
community. Overall, our work highlights that if both these services continue to increase they
will become part of the web’s critical infrastructure with profound impact in it’s usage and
performance, creating a need for the study and development of alternative architectures and
distribution methods.
|