PastPresentFuture The Past, Present and Future of P2P (Peer-to-Peer) Ming-Wei (Benson) Wu Dept. of Electrical Engineering National Taiwan University & Taiwan Network Information Center (TWNIC) 2003/08
National Taiwan University Copyright Reserved Ming-Wei (Benson) Wu Outline I.Past 1. P2P basics (How everything started?) 2. P2P taxonomy and vendors (Who’s Who) 4. P2P trend in Taiwan (MP3 file-sharing) II.Present 1. P2P problems (connectivity, scalability and security) 2. P2P solutions for the mentioned problems 3. P2P business models in Taiwan (EZPeer and Kuro) III.Future 1. P2P itself 2. P2P and others 3. P2P impacts
National Taiwan University Copyright Reserved Ming-Wei (Benson) Wu P2P basics How everything started? First, what Internet people want? When they are still a newbie…they want to “join” …Client-to-Server When they become big enough…they want to “share” …Peer-to-Peer P2P = Peer-to-Peer Equal in existence “We talk to each other directly”, no middleman. Equal in role “I do what you can do”, no bias.
National Taiwan University Copyright Reserved Ming-Wei (Benson) Wu P2P Taxonomy and Vendors So…what do we do? For better availability (where is it?) Data/Information sharing (one data, many copies) File-sharing: all kinds of file (mp3, video, picture, documents, etc) e.g Gnutella, E-Donkey, KaZaa, Mojo Nation, WinMX For better productivity (do it together!) Data/Information collaboration (many people, one project) Groupware: many people working together on one project e.g. Groove (Lotus), E-Speack (HP), Jabber For better utilization (no one should be “idle”) Resource sharing (I borrow your CPU time and you borrow my disk space) P2P computing: Leveraging of CPU, memory, storage resources e.g. Entropia, Legion (Applied MetaComputing)
National Taiwan University Copyright Reserved Ming-Wei (Benson) Wu More P2P Vendors Collaboration : Edge Service : Intelligent-agent : File-sharing : Distributed Computation :
National Taiwan University Copyright Reserved Ming-Wei (Benson) Wu P2P trend in Taiwan File-sharing? Yes. Namely “Kuro” and “EZPeer” Groupware? Not really. So far still using Lotus Notes. No global communications. Very little Groove (Lotus). P2P computing?: No.
National Taiwan University Copyright Reserved Ming-Wei (Benson) Wu P2P Problems Connectivity problem I cannot talk to peers whom are behind a NAT/Firewall Especially opposing NATs/Firewalls Scalability problem How big? Can this P2P network grow? How fast? To locate my targeted resources? Security problem How secure? (file-sharing as an example) Correct name, wrong content (malicious file content) e.g. File name “A” actually has content “B” in it.
National Taiwan University Copyright Reserved Ming-Wei (Benson) Wu P2P Solutions Connectivity solution (quite ad-hoc, quite temporal) Use a middleman to help We all talk to the middleman first Scalability solution (good) How big? Use smart message routing, never use flooding/broadcasting e.g. Very much like talking in a room. If everyone talks through shouting The room cannot afford too many people (too noisy) How fast? Use distributed hashed table (DHT), indexed everything e.g. Very much like Google search engine indexed every single web page on the Internet Security solution (good) How secure? (file-sharing as an example) Use file checksum/digest (known as the DNA of a file) e.g Very much like we check MD5 or PGP key of a downloaded file
National Taiwan University Copyright Reserved Ming-Wei (Benson) Wu P2P Business Models in Taiwan MP3 file-sharing service (two biggest: Kuro and EZPeer) Charge: 3 US dollars per month (unlimited download) Members: around 0.5 million users (500,000) Source of MP3: from its users Storage of MP3 list: a directory server (maintained by Kuro) Storage of MP3 files: its users’ computer License of MP3: None Impact: IFPI-Taiwan (a representative of music industry in Taiwan) estimates about 250 millions of US dollars (250,000,000) are lost due to MP3 downloading in Taiwan Results: IFPI-Taiwan is very mad and is taking legal action against Kuro to EZPeer.
National Taiwan University Copyright Reserved Ming-Wei (Benson) Wu An example of a middleman scenario– Relay peer of JXTA
National Taiwan University Copyright Reserved Ming-Wei (Benson) Wu P2P Future: P2P Itself Location-independent Everywhere (e.g. Home, office, school, etc) Device-independent any-where (e.g. PDA, PC, Notebook, TabletPC, etc) Always-on/Always-available any-time (e.g. 24/7/365) P2P will be pervasive.
National Taiwan University Copyright Reserved Ming-Wei (Benson) Wu P2P Future: P2P and Others P2P makes differences in your life MP3 downloading, ICQ/MSN messenger, Online games Long-existed client-to-server services Some services (e.g. web browsing) don’t need P2P features …old bones won’t die, just faded. Up-and-coming Web Services Some services (e.g. travel agency) needs one-stop shopping (including tickets, hotel reservation, cars, etc) …a need for sever-to-server is rising!
National Taiwan University Copyright Reserved Ming-Wei (Benson) Wu P2P impacts as a Conclusion [Relationship?] More close human interactions in CyberSpace From shaking hands to clicking mouse [Life?] More realization of real-world-like activities in CyberSpace From walking to browsing [Business?] More integrations, less pop-up From one-at-a-time to kill-them-all Umm…lazybones?
National Taiwan University Copyright Reserved Ming-Wei (Benson) Wu Thank you~ Q&A