Transcript Networking in 5-50 Years

Networking in 5-50 Years –
applications and
requirements
Henning Schulzrinne
Columbia University
[email protected]
29-Mar-16
Future of Networking
Overview
 Infrastructure, once established, tends to change
very slowly
 Hypothesis:
 all major communications modes have been explored
 replacement dedicated  IP largely complete
 Networking lacks obvious drivers of other
technologies:
 energy costs, pollution  fuel cells
 higher speed  jet engine
 New applications not necessarily bandwidth-driven,
but cost-driven
 Reliability is the real QoS
29-Mar-16
Future of Networking
Networking is getting into
middle years
29-Mar-16
idea
current
IP
1969,
1980?
1981
TCP
telnet
ftp
1974
1969
1980
1981
1983
1985
Future of Networking
Standardization

Really two facets of standardization:
1. public, interoperable description of
protocol, but possibly many (Tanenbaum)
2. reduction to 1-3 common technologies



LAN: Arcnet, tokenring, ATM, FDDI, DQDB, …
 Ethernet
WAN: IP, X.25, OSI  IP
Have reached phase 2 in most cases,
with RPC (SOAP) and presentation
layer (XML) most recent 'conversions'
29-Mar-16
Future of Networking
Technologies at ~30 years
 Other technologies at similar maturity
level:
 air planes: 1903 – 1938 (Stratoliner)
 cars: 1876 – 1908 (Model T)
 analog telephones: 1876 – 1915
(transcontinental telephone)
 railroad: 1800s -- ?
29-Mar-16
Future of Networking
Observations on progress
 1960s: military  professional  consumer
 now, often reversed
 Oscillate: convergence  divergence
 continued convergence clearly at physical layer
 niches larger  support separate networks
 Communications technologies rarely
disappear (as long as operational cost is low):
 exceptions:
 telex, telegram, semaphores  fax, email
 X.25 + OSI, X.400  IP, SMTP
 analog cell phones
29-Mar-16
Future of Networking
History of networking
 History of networking = non-network
applications migrate
 postal & intracompany mail, fax  email,
IM
 broadcast: TV, radio
 interactive voice/video communication 
VoIP
 information access  web, P2P
 disk access  iSCSI, Fiberchannel-over-IP
29-Mar-16
Future of Networking
Network evolution
 Only three modes, now thoroughly explored:
 packet/cell-based
 message-based (application data units)
 session-based (circuits)
 Replace specialized networks
 left to do: embedded systems
need cost(CPU + network) < $10
cars
industrial (manufacturing) control
commercial buildings (lighting, HVAC, security; now
LONworks)
 remote controls, light switches
 keys replaced by biometrics




29-Mar-16
Future of Networking
New applications
 New bandwidth-intensive applications
 Reality-based networking
 (security) cameras
 Distributed games often require only lowbandwidth control information
 current game traffic ~ VoIP
 Computation vs. storage vs. communications
 communications cost has decreased less rapidly
than storage costs
29-Mar-16
Future of Networking
Commercial access cost (T1)
$700
$600
$/month
$500
$400
$300
$200
$100
$0
1996
1998
2000
2001
Year
29-Mar-16
Future of Networking
T1
2002
2003
Transit cost (OC-3, NY –
London)
29-Mar-16
Future of Networking
Disk storage cost (IDE)
Cost
$100,000.00
$/GB
$10,000.00
$1,000.00
$100.00
$10.00
$1.00
May-79
Feb-82
Nov-84
Aug-87
May-90
Jan-93
Date
29-Mar-16
Future of Networking
Oct-95
Jul-98
Apr-01
Jan-04
Transition of networking
 Maturity  cost dominates
 can get any number of bits anywhere, but
at considerable cost and complexity
 casually usable bit density still very low
 Specialized  commodity
 OPEX (= people) dominates
 installed and run by 'amateurs'
 need low complexity, high reliability
29-Mar-16
Future of Networking
Security challenges
 DOS, security attacks  permissions-based
communications
 only allow modest rates without asking
 effectively, back to circuit-switched
 Higher-level security services  more
application-layer access via gateways,
proxies, …
 User identity
 problem is not availability, but rather overabundance
29-Mar-16
Future of Networking
Scaling
 Scaling is only backbone problem
 Depends on network evolution:
 continuing addition of AS to flat space 
deep trouble
 additional hierarchy
29-Mar-16
Future of Networking
QoS
 QoS is meaningless to users
 care about service availability 
reliability
 as more and more value depends on
network services, can't afford random
downtimes
29-Mar-16
Future of Networking
Wildcards
 Quantum computing
 Teleportation
29-Mar-16
Future of Networking