Transcript History of P2P

®
A Brief History of P2P Networks
Laird Popkin, CTO, Pando Networks
Goal of this Presentation”
The goal of this presentation is to present a quick
overview of p2p technology so that we’re all talking the
same language.
Please jump in with questions, corrections, etc.
The Internet is “Peer to Peer”
The fundamental technology of the Internet is TCP/IP. At this level, the
internet composed entirely of “peers” connected by a mesh network.
Everyone had an internet IP address, and could be both a client and a
server. This allows the internet to scale “infinitely”.
But client/server applications (FTP, HTTP) are much easier to implement.
This creates many independent “islands”, each with a scaling problem if
they’re popular. Witness the “Slashdot Effect”.
With P2P, the more popular something is, the more powerful it becomes,
matching the fundamental scalability of the Internet. This requires more
cleverness than client/server, but that just makes it more fun.
P2P Generations”
P2P used to mean “file sharing”. For example:
•
First Generation: Napster
•
Second Generation: Kazaa, Gnutella, eDonkey
•
Third Generation: BitTorrent
Now: P2P means many different things
First Generation: Napster
QuickTime™ and a
TIFF (LZW) decompres sor
are needed to see this picture.
The first popular p2p application
was Napster. Launched in 1999, it
collected an index of all of the
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
music files on users’ computers
and provided a centralized
searchable database, then
retrieved the music from one
users’ computer.
Worked great. Sued immediately,
shut down in 2001.
Diagram from HowStuffWorks.com
Second Generation - Decentralization: Kazaa
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
Decentralized architecture, computers form
a “mesh” that performs searching and file
delivery.
–
Added “supernodes” to coordinate
communication.
–
Bundled in adware, spyware, etc., to
monetize installs.
•
Trained people to avoid p2p apps.
•
Lead to “hacks” such as KaZaa Lite
–
Weak security
•
–
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
Lead to interference in the network
Upgrade server was single point of control
(used to shut off Morpheus).
Was extremely popular, until lawsuit.
Diagram from http://cis.poly.edu/~ross/papers/UnderstandingKaZaA.pdf
Understanding KaZaa from Brooklyn Polytechnic
Second Generation - Open: Gnutella
QuickTime™ and a
TIFF (L ZW) d eco mpres sor
are nee ded to s ee this picture.
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
Decentralized architecture, computers form
a “mesh” that performs searching and file
delivery.
–
No central database (so can’t be shut down
by a lawsuit)
–
Open protocol (many clients)
–
Completely decentralized is very hard to
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
make work.
•
“bootstrap problem”: how do new clients join a
“decentralized” network?
•
Cascade problem: searches can crush network.
Morpheus adopted, but Gnutella didn’t
scale well, so growth was limited.
Gnutella community of developers has
continued to evolve and grow. Added
ultrapeers, Bitzi lookups, etc.
Diagram from
Limewire.com
Second Generation - Swarm: eDonkey
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
eDonkey introduced:
–
Magnet links, allowed creation of web sites
–
Swarm delivery extremely fast
–
Protocol was reverse engineered, allowing
creation of eMule, etc.
–
Was extremely popular, until shut down.
–
Architecture wasn’t fully distributed, so
shutting down servers shut down network.
Attempted to migrate to fully distributed Overnet
Diagram from http://www.ed2k-serverboard.de/diesel/edonkey/Serverbeschreibung.html
Third Generation - Secure, Swarming: BitTorrent
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
Secure, Swarm Delivery has many
Diagram from Bram’s 2002 white paper
“Incentives Build Robustness in BitTorrent”
advantages.
–
Resilient to communication errors and
interference.
–
Can be extremely fast for popular content.
–
Tit-for-tat penalizes “cheaters”
–
Tracker makes each torrent independent,
allows for overall network to scale,
–
No search avoids legal risks
Also:
–
Open protocol, many implementations
–
Widely adopted, probably over 50% of
internet by data volume
–
Web links lead to “tracker” web sites. Are web
sites in Norway “safe”?
–
DHT’s remove dependence on Trackers
Now P2P is growing in many directions
P2P is exploding in range. Some examples:
•
AllPeers integrated BitTorrent into FireFox for social file sharing
•
BitTorrent pursuing eCommerce, technology licensing
•
CacheLogic is providing a P2P CDN, with variable QoS/pricing.
•
Distributed Computing Industry Association facilitates policies, education, etc.
•
Democracy Player subscribes to RSS+BitTorrent
•
Joost and Babelgum use P2P to provide a “TV Experience” over the internet
•
Pando made sending easy, integrated into user applications, video platform.
•
PeerApp and Oversi provide caching servers to ISP’s
•
PPLive, PPStream, etc., focusing on live video streaming
•
Shareaza, Xfactor, etc., provide a common GUI over multiple protocols.
•
Solid State and One Click Media are browser plug-ins that play “stream on demand” video
•
Vuze, was Azureus, now focused on video download/playback
•
Who did I forget?
®