Transcript Growing Pains: The Internet in Adolescence

Growing Pains:
The Internet in Adolescence
Fred Baker
ISOC Chairman of the Board
Cisco Fellow
The parable of the swing
Today’s Internet

The optical internet
backbone
Gigabit to terabit links

Campus
Networks
(LANs)
Internet
in Airlines

Access networks
xDSL, cable modem, ISDN,
asynchronous dial
20,000 instantaneous sessions per GBPS
backbone bandwidth

UNIVERSITY
Brief History of the Internet
Comic Book to Cyberspace
Datagram Switching

Len Kleinrock, 1962



The strength of a chain
is its weakest link
The strength of a web is
its surviving path
Datagram Switching


Developed at
UCLA+Xerox PARC
DARPA Funding
Early commercialization
Source: http://www.cidr-report.org
Killer Applications: Early Business
Borderless
Mail, FTP, Archie,
Adoption
Business
Network News
Consumer
Multi-player
WWW, IRC
Adoption
Games
Projected routing table
growth without CIDR/NAT
Moore’s Law and NATs,
with aggressive address
conservation policy, make
routing work today
Deployment Period of CIDR
Marketing rushes in where
engineering fears to tread

Internet bubble:



“Build it and they will come”
“New Economy” where profitability is
irrelevant
“.com” era
Profitability…


…The Final Frontier.
Companies are operating on the
premise that if it doesn’t make money,
it is not a good business to be in…
Status of Internet Technology in
developed nations

A utility:





Water, Sewer
Electricity, Natural Gas
Telephone
Internet
Internet access and facility is assumed
in education, business, and increasingly
in society
The Digital Divide

“In addressing the digital divide, Uganda and
other countries in the region face three broad
challenges:




Creating and exploiting access to external
information resources;
Creating internal information resources;
Creating and exploiting access to internal
information resources.
A common underlying factor that cuts across
the three broad challenges is the need for a
competent human resource.”
Dr. F. F. Tusubira
Makerere University, February 2003
Client/Server Architecture is
overtaken by events

For web:

Sufficient to have clients
in private address spaces
access servers in global
address space
Private
Address
Realm
Private
Address
Realm
Global
Addressing
Realm

Telephones/Point to Point

Need an address when you call
them, and are therefore servers
in private realm
Who are today’s application
innovators?

Open Source example: Freenet/KaZaA






Large-scale peer-to-peer network
Pools the power of member computers
Create a massive virtual information store
Open to anyone
Highly survivable, private, secure, efficient,
http://www.firenze.linux.it/~marcoc/index.php?page=w
hatis
History of the IETF
Originally supporting Research
Networks

Dates:




Started 1986
Non-US participation by 1988
First non-US meeting: Vancouver, August 1990
Constituents:





Originally US Government only
Added NSFNET (NRN), education, research
Eventually added vendors
The government left…
International participation
Characterizing the community:

Semi-homogenous


“Netiquette”


People largely knew and trusted each other
Anti-social behavior drew direct and public
censure as “impolite”
Key interest:

Making the Internet interesting and useful for
themselves and their friends.
IETF Mission Statement

Make the Internet Work


Whatever it takes…
But what is the Internet?


IPv4? IPv6? MPLS?
Applications like WWW? Mail? VoIP?
Historical principles





End to End principle
Robustness principle
Rough Consensus
and Running Code
Institutionalized
altruism
Mutual Benefit






Managed Trust
Highly relational
Principle of least
surprise
Openness
Anti-kings
Achieving “right”
results because they
are right
Now supporting all IP-based
Networks

Constituents:


Researchers
Network Operators


Implementers (engineers, often from vendors)



ISP, NRN, Enterprise
Large percentage of attendees
Interactions with various governments…
Fully international participation
Characterizing the community:

Heterogeneous


“Netiquette”



Business reasons for involvement
Expectation of safe environment
Moving towards codification of
expectations
Key interest:

Defining technology to use or to sell
Undercurrents

Business agenda



Intellectual Property Issues


Business relationships rather than personal
relationships
Political process
About protecting ideas, not sharing them
Civil servants as leaders
IETF: in a maze of twisty
passages – all different
What makes IETF hard?
Breakdown of trust


Community sees leaders as a cabal
Leaders see community that designs for
narrow scope of applicability or misses
key issues
What makes IETF hard?
Opaque processes





RFC Editor
Secretariat
Internet Assigned Number Authority
Internet Engineering Steering Group
Internet Architecture Board
What makes IETF hard?
Consensus process


Lack of comment interpreted as
consent, but may mean loss of interest
Consensus may not be desired by
participants seeking market share
What makes IETF hard?
Personal responsibility

Expectation that “they” should do
something:


IETF composed of people, and people do
the work
Personal involvement essential to progress
The IESG is rapidly
approaching a solution
Sounds like bad news

Not really



The IETF is just deciding what it wants to
be when it grows up…
Quite a bit of good work going on there
Other groups of interest


NANOG, Apricot, RIPE, etc
Many others
What is next for the Internet?
High-end research backbones

Combining IP routing and optical
routing in overlay networks



“Designer networks” for research purposes
Production networks for applications
What parts of network to research?



Routing (IP, Optical)
Applications
IPv6-based
GARDEN
Network Topology
NTT via NYI to SuperSINET *)
SE
Edmonton Alberta
COP
*)
UKLight *)
CANARIE
1GE to 10GE
CHI
T-Systems
To US
LON
NetherLight
AMS
StarLight
Chicago
Global Crossing
POZ
DANTE POP
KRA
NYC
FRA
PAR
CERN
MIL
Dark Fiber / Lambda / 10G
PRA
VIE
BUD
SurfNet / 10G
CESNET / 2,5G
LambdaNet / 2,5G
Nordic Connections
*)
under discussion
Ukerna / 10G
via GEANT / 2.5G
via SWITCH / 2.5G
ATH
High Speed Optical Domains
GARDEN
Project Structure
Project Management
WP0
Advanced
Protocol
& Service
Deployment
WP2
Protocol
Measurement
Mgmt
&
Security
&
Architecture
& AAA
Provisioning
Research
WP4
WP5
WP3
Integrated IP + Optical
Network
WP1
Technical Support, Dissemination,
Training and Demonstration
WP7
High Bandwidth
Real Time Applications
WP6
10GE
Production Network
Optical
STM-64/OC-192
STM-16/OC-48
Production
GE
Research
建議電路
中央研究院
台灣大學
東華大學
Taipei
C7609
C7609
中央大學 C7609
C7609
GSR
中正大學
交通大學
C7609
C7609
成功大學
TWAREN
GSR
新竹
C7609
GSR
Tainan
Hsin-chu
中興大學
清華大學
C7609
暨南大學
GSR
中山大學
C7609
C7609
Taichung
C7609
10GE
Research Network
台灣大學
Taipei
C7609
Optical
STM-64/OC-192
STM-16/OC-48
Production
GE
Research
建議電路
中央研究院
C7609
GSR
東華大學
C7609
ONS15600
ONS15454
ONS15454
中央大學
ONS15454
C7609
TWAREN
Hsin-chu
Tainan
ONS15600
交通大學
C7609
中正大學
C7609
ONS15600
GSR
GSR
成功大學
C7609
ONS15454
清華大學
C7609
ONS15454
GSR
中興大學
C7609
ONS15454
Taichung
暨南大學
C7609
中山大學
C7609
10GE
Optical Network -1
STM-16/OC-48
Production
Taipei
(2)
GE
Research
(#)
中央研究院
C7609
台灣大學
C7609
STM-64/OC-192
Optical
建議電路
電路數量
東華大學
(2)
GSR
ONS15454
C7609
(2)
ONS15454
ONS15454
(2)
(6)
中正大學
中央大學
C7609
ONS15600
C7609
ONS15454
(2)
ONS15454
(2)
ONS15454
TWAREN
ONS15454
交通大學
C7609
ONS15454
(6)
(3)
(2)
Hsin-chu
ONS15454
ONS15600
ONS15600
(3)
(6)
Tainan
成功大學
C7609
(2)
GSR
GSR
ONS15454
ONS15454
(4)
清華大學
C7609
(2)
ONS15454
中興大學
C7609
(2)
ONS15454
(2)
暨南大學
GSR
Taichung
C7609
(2)
ONS15454
中山大學
C7609
(2)
ONS15454
Proposed UN-FAO “Growing
Connection”: Ghana
384 KBPS
Or E1
Internet
Long distance
IEEE 802.11
Database.library.de
Village.school.gh
several PCs + Router
Village.school.gh
several PCs + Router
Village.school.gh
several PCs + Router
42
Manet looks at a mobile
infrastructure


“Enterprise” infrastructure network
 Connects roaming devices which
themselves form the infrastructure
 Neighbor relationships change
randomly in routing
 Not appropriate as backbone
Fundamental issue:
 Not “can I find the addressed
device/prefix in my network”, but
 “Is there a usable route to the
addressed device/prefix.”
43
Today’s Client/Server access
control


We trust people to access
servers and do limited
operations on them
As a result, we limit our
applications by the power
of the servers we run
them on
44
Peer-peer access control
model

Let everyone talk



Distributed computing
Peer computers to perform
function, not server
Central Authentication/
Authorization


Access control
Accountability
45
What needs to change?

Effective prophylactic security



Firewall ≠ Network Address Translator
Secure Firewall Traversal
Secure identity/authority management



Spam management…
Good point-to-point application software
and models (Freenet/KaZaA?)
Managability…
“As new IP communications services and
devices become available, they may
stimulate new demand and increase VoIP
traffic flows beyond the growth rates
characteristic of the traditional voice
telephony market.
… the total market may reach … six percent
of the world's forecasted international
traffic for the calendar year 2001”
Telegeography 2002
47
Voice/Video on IP networks
Billing/
Authorization
Control
Plane
Data
Path
Video on Demand…
Video-on-demand
Server located in
the POP
Internet Router
located in the POP
100-baseT to Home
Carrying multiple
Video streams plus
Voice and data
Forensics in an Internet environment


Who did they “speak” with?
What did they “say”?
IP Control
Traffic
Control Device:
Call Manager, SIP Proxy,
Authentication Server, etc
Log
Stream
Control
Mediation
Warrant
Intercept
Configuration
Intercepted
Information
Data ACL
Intercepted
Data
Data Mediation
IP Data
Growing Up…





Profitability…
User-tolerant (if not friendly)
applications
Business-tolerant applications…
Manageable applications and networks
Convergence…
Growing Pains:
The Internet in Adolescence
Fred Baker
ISOC Chairman of the Board
Cisco Fellow