[talks] J Nelson generals
Melissa M. Lawson
mml at CS.Princeton.EDU
Wed May 9 14:48:09 EDT 2012
Jude Nelson will present his research seminar/general exam on
Wednesday May 16 at 10AM in Room 402. The members of his
committee are: Larry Peterson (advisor), Michael Freedman,
and Vivek Pai. Everyone is invited to attend his talk, and
those faculty wishing to remain for the oral exam following
are welcome to do so. His abstract and reading list follow
On the Internet today, an increasingly common strategy for efficiently disseminating a large dataset to a scalable number of readers (but not writers) is to employ a content distribution network (CDN). Readers query the CDN for the data, and the CDN reads the data from its origin servers and replicates the data in its network caches, thereby allowing the caches to service future read requests. However, there are two limitations in this approach that hinder
origin servers and clients from achieving read/write semantics on the dataset. First, a CDN's caches are neither coherent nor consistent--if the data on an origin server changes, there will be a window of time where the CDN may
still cache and serve stale or even corrupt data. Second, each CDN implements its own content naming scheme and caching policy, requiring origin servers and possibly clients to be designed for a specific CDN instance.
We address these limitations with Syndicate, a wide-area read-write distributed filesystem that transparently leverages arbitrary CDNs to drive data distribution between its clients. A mounted Syndicate filesystem exposes a client's files to the CDN and translates other clients' remote read operations into CDN-specific data queries. Regardless of the CDN's caching policies, Syndicate ensures that readers only read the latest known versions of each file, and ensures that each file has close-to-open consistency in the face of multiple writers. Syndicate provides clients the means to coordinate file discovery and I/O through a fault-tolerant metadata service, as well as the means
to back up their files to arbitrary remote storage through a global replication service. By leveraging CDNs to handle a scalable number of readers and allowing clients to fully control file and metadata propagation, Syndicate fully decouples the problem of file persistence from read performance--Syndicate clients need only concern themselves with enhancing the persistence of their data, beyond presenting it to applications.
In this work, we will present Syndicate's architecture, protocols, and design. We will then present a series of micro and macro benchmarks of our Syndicate prototype, which is currently deployed on PlanetLab and VICCI and leverages a private deployment of the CoBlitz CDN.
Peterson, Larry L., and Davie, Bruce S. "Computer Networks, a Systems Approach. 4th Ed."
Cerf, V., and Kahn, R. "A Protocol for Packet Network Intercommunication."
Lamport, L. "Time, clocks, and the ordering of events in a distributed system."
Schneider, F. "Implementing Fault-Tolerant Services Using the State Machine Approach: A Tutorial"
Park, K. and Pai, V. "Scale and Performance in the CoBlitz Large-File Distribution Service."
L. Wang, V. Pai, and Peterson, L. "The Effectiveness of Request Redirection on CDN Robustness."
Annapureddy, S., Freedman, M., and Mazieres, D. "Shark : Scaling File Servers via Cooperative Caching."
Burrows, M. "The Chubby lock service for loosely-coupled distributed systems."
Peterson, L., Bavier, A., Fiuczynski, M., Muir, S. "Experineces Building PlanetLab."
Belaramani, N., Dahlin, M., Gao, L., et. al. "PRACTI Replication."
Stribling, J., Sovran, Y., Zhang, I., et. al. "Flexible , Wide-Area Storage for Distributed Systems with WheelFS."
Freedman, M., Freudenthal, E., Mazieres, D. "Democratizing content publication with Coral."
Vahdat, A., Eastham, P., Yoshikawa, C., et. al. "WebOS: Operating System Services for Wide Area Applications."
More information about the talks