Transcript Peer to Peer

Comparing Peer to Peer
File Sharing Technologies
David Strom, [email protected]
O’Reilly P2P DC Conference
Sept 2001
(c) David Strom Inc. 2001
1
Summary
•
•
•
•
•
•
Brief history of P2P file sharing
The Post-Napster world
Technology examples
Business models
Motivations for using P2P file sharing
Corporate strategies
(c) David Strom Inc. 2001
2
Brief history of file sharing
• Non-music related
• Music-related
(c) David Strom Inc. 2001
3
Non-music related
•
•
•
•
Set up a file server on Netware, Unix, NT
Set up FTP server on the Internet
Upload files to various newsgroups
Send email around with attachments
(c) David Strom Inc. 2001
4
Music-related FS
• First came used record stores
• Copy a CD to tapes
• Then came MP3s
(c) David Strom Inc. 2001
5
Remember MP3 downloads?
•
•
•
•
Locate an ftp server with music on it
Try to connect to the server (often busy)
Upload a few songs first
Then hopefully find the right song and
download it
• And much more work to host your own
music ftp server!
(c) David Strom Inc. 2001
6
Music-related FS (con’t)
• MP3s were quickly followed by busy ftp
servers and ratios
• Then Napster
• Quickly followed by lots o’ lawyers
• Other transports and networks, like AOL IM
(Aimster)
• Non-centralized directories (Gnutella)
(c) David Strom Inc. 2001
7
Problems with music FS solved
by Napster
• ftp sites often busy
• Had to learn the ins and outs of ftp client
software
• Uploading requirement was onerous
• Search tools crude, could often only find ftp
sites, not songs
• Setting up servers painful too.
(c) David Strom Inc. 2001
8
Post-Napster world
•
•
•
•
Gnutella clients (LimeWire, BearShare)
FastTrack clients (MusicCity, KaZaa)
Groove
Others that don’t have anything to do with
music and pre-teens
• RIAA non-starters like SDMA
(c) David Strom Inc. 2001
9
Silly question
If the music companies could figure out a
technology that could allow reasonable persong payments (like 25 cents), would all
this industry ever happened?
(c) David Strom Inc. 2001
10
But the real killer P2P FS app:
• PORN!
• Porndigger is one example
• KaZaa et al. allow image/video searches!
(c) David Strom Inc. 2001
11
Comparing KaZaa and LimeWire
•
•
•
•
•
•
L: Multiple concurrent searches
K: Better and faster results
L: more users, K: more files
K: better traffic/path analysis
L: Java sluggishness in UI
Neither will replace Napster!
(c) David Strom Inc. 2001
12
Common elements of all P2P FS
technologies
•
•
•
•
Easy to search and share files
Uses http and ftp protocols
Each user's PC acts as client and server
A community of users with common goals
and interests -- maybe
• Build upon P2P services to increase the
overall value of the system
(c) David Strom Inc. 2001
13
The metrics for P2P FS
• Searchable directory
• Monitor/control incoming/outgoing traffic
• Access controls and scope of file sharing on
each PC’s disk
• Speed of locating other peers and keeping
connection up
(c) David Strom Inc. 2001
14
How distributed does your
network have to be?
• Napster: central server to authenticate and
track users
• Gnutella: everything is distributed
• Aimster: uses AOL IM directory to
authenticate/track users
(c) David Strom Inc. 2001
15
But the real reason these services
are proliferating:
• I want your (music, porn, whatever) files!
• Sharing my files with you isn’t a real
motivation – yet.
(c) David Strom Inc. 2001
16
Shirky’s list of lessons learned
from Napster
• Centralize what you need to make your
business model work
• A poor user interface isn’t an issue,
especially if people can still get things done
• It routes around network admins/firewalls
• It allows people to create their own
namespaces without specialized knowledge
(c) David Strom Inc. 2001
17
Review these concepts for new
P2P FS technologies
• Centralize nothing so lawyers can’t get you
• UI will become more important as
competition increases
• Routing around Internet blockades is
important
• The shared collection is critical if the
community is going to thrive
(c) David Strom Inc. 2001
18
Let’s look at non-P2P FS
• Mydocsonline or similar “Internet hard
disk”
• PGP secure email, maybe
• Quickbase.com
• Maybe even a web-based Office suite like
TeamOn, FreeDesk or Blox
(c) David Strom Inc. 2001
19
Problems
• Motley collection difficult to manage
• Not everyone is willing to work with
multiple tools
• Authentication is difficult – multiple logins
• Still need to download lots of software,
despite “100 % browser” emphasis
• Tiresome email exchanges of documents
• Security by obscurity
(c) David Strom Inc. 2001
20
One contender: Groove Networks
• Authentication is always on and setup is easy
• Different workspaces for different collections
of people
• No browser, email file swapping needed
• Supports a wide variety of tools and
information sharing models
• Great for sharing files between home/work PC
(c) David Strom Inc. 2001
21
Other aspects of Groove
• Two-way sharing model vs. sons-ofNapster steal-this-file approach
• Not clogged with people looking to
download music or porn files, yet
• May be overkill for simple home/work
file sharing, though
(c) David Strom Inc. 2001
22
What business models make
sense?
•
•
•
•
•
Advertising
User charges
Corporate site licenses
Core technology licenses
ISP/ASP volume purchases
(c) David Strom Inc. 2001
23
Motivations: the network edge
• Right now network edge devices are underused or
dumb when it comes to working with the network
fabric itself
• Edge is more reliable than core servers from
application perspective (sometimes)
• P2P FS adds intelligence, increased utilization at
the edge
• Makes it easier for people to get more work done,
just like c. 1981 PCs
(c) David Strom Inc. 2001
24
P2P FS creates new kinds of
“edge-to-edge” networks
• Only need a portion of the Internet
infrastructure to be working
• Really targeted at small collections of
people that need intense collaboration
• My desktop needs to talk to your desktop
• Need to get around firewalls, VPNs, and
other corporate protection
(c) David Strom Inc. 2001
25
Controlling your own network
destiny
• Ask your IT department for the capability to
bring up a new file server in the next hour
without prior budget approval
• Now imagine being laughed out of the room
(c) David Strom Inc. 2001
26
Advantages of P2P FS
• No need for setting up IP addresses, DNS
configuration, firewall rules, web server
hosting co-location agreements, IT policy
pronouncements, lengthy approvals up the
IT command structure, IT R&D review, IT
rollouts, …
• Do you detect a pattern here?
(c) David Strom Inc. 2001
27
Drawbacks to P2P FS
• Security loopholes
• Loss of control over desktop activities
• Some companies don’t like to share and
play with others
(c) David Strom Inc. 2001
28
Corporate strategies
• Download and try Groove or uRoam
• Compare with downloading music or
PC/Anywhere for functionality
• Examine your own programs for file transfer
components that could benefit from peering
• Examine whether you can benefit from shared
processing models
• Setup a trial Groove workgroup and project team
(c) David Strom Inc. 2001
29
Places for more information
• Copies of this presentation
strom.com/pubwork/P2P9-01.ppt
• My review of Groove:
strom.com/pubwork/varbiz.html
• List of P2P FS technologies:
4arrow.com/users/my/html/sections.php?
op=viewarticle&artid=204
(c) David Strom Inc. 2001
30