Speaker
Jeremy Stribling
(MIT CSAIL)
Description
It is a challenge to build applications that need to share
data and are distributed across hundreds or thousands of
computers in a wide-area network (e.g., PlanetLab or on a
Grid). In order to cope with high latency, throughput
bottlenecks, and temporary failures, such applications
typically implement their own storage plan or use
special-purpose storage solutions (e.g., DISC, Globus,
DHash, etc.). Inspired by the success of the Google File
System for cluster applications, we investigate whether a
general-purpose wide-area file system could simplify
building distributed applications. In particular, this talk
presents WheelFS, a new distributed file system that eases
the development of distributed applications such as
cooperative Web caches, data-intensive Grid applications,
and distributed email systems. By giving applications a
familiar POSIX interface, and giving them control of how the
file system handles their data in the face of the
challenging wide-area network, WheelFS simplifies the
development of new applications and allows reuse of much
existing code -- often distributed applications can be
created with just a few changes to a configuration file.
This talk will present the design and implementation of
WheelFS, and evaluate its performance in the context of
several applications.