I read some eye-opening stuff this weekend.
1. Have you ever heard of CoralCDN? I hadn't. But Digg, SlashDot and Fark.com use it to serve links. And about 300 other sites have "Coralized" their software, image, and audio/video downloads. Coral is a P2P content distribution network run by New York University.
Both end users and site owners can take advantage of the free service by appending ".nyud.net:8080" to any URL. Once you access any Coralized content, your computer automatically becomes a distribution node. Coral's distributed sloppy hash table (DSHT) indexing system creates self-organizing clusters of nodes that communicate with each other to optimize download speed and avoid hot spots in the infrastructure.
And that's not all. Like other large websites and content distribution networks, CoralDNS sends requests to the "closest" node based on the client's public IP. But this approach does not take into account the distance between clients and their DNS resolvers. Earlier this year Coral launched the Illuminati network measurement project to quantify this distance, as well as explore the extent of NAT and proxy usage. The results could help refine CoralDNS' algorithm.
2. Through CoralDNS' website, I found my way to an MIT paper on the Cooperative File System (CFS). It's a P2P read-only storage system that runs on Linux, OpenBSD, and FreeBSD. CFS' decentralized architecture uses a distributed hash table for block storage. Blocks are cached and replicated to optimize speed and redundancy. With 4,096 servers, retrieving a block of data involved contacting only seven servers. Performance was unimpaired even when as many as half the servers failed.
3. The website for Berkeley's somewhat related OceanStore project was last updated in 2002. Even then, researchers were working on a prototype for a high availability storage utility than runs atop an infrastructure of untrusted servers.
Any computer can join the infrastructure, contributing storage or providing local user access in exchange for economic compensation. Users need only subscribe to a single OceanStore service provider, although they may consume storage and bandwidth from many different providers. The providers automatically buy and sell capacity and coverage among themselves, transparently to the users. The utility model thus combines the resources from federated systems to provide a quality of service higher than that achievable by any single company.
4. Last but not least, there's Pastry, whose project members include Microsoft Research, Purdue and Washington University. It's an application-indepedent overlay network for routing requests and locating data among P2P nodes. It includes mechanisms that support application-specific object replication, caching, and fault recovery.
None of these ideas have yet been commercialized, but don't you think it's only a matter of time?
Yesterday, in response to a New York Times article on voice over WiFi, Tom Evslin predicated that
WiFi support in mobile phones will shift the balance of power from the big wireless operators to the cellphone hardware and software makers. Phones will be purchased independently of calling plans just as computers are purchased independent of Internet connectivity arrangements. Coupons for access may be included with phones instead of phones being included with calling plans. Why? Because voice calling will be too cheap to meter and hardware will still cost something.
Might the relationship between site owners and web hosting providers undergo the same transformation as a result of P2P technologies?
Comments