Transcript papers 1 and 2 - University of California, Berkeley

University of California/Berkeley
Internet and IPv6
Reviews for EE290T
Minghua Chen
[email protected]
University of California/Berkeley
Outline
• Internet – “The Internet: a tutorial”, by J.
Crowcroft
• IPv6 – “The next generation of the
Internet: aspects of the Internet protocol
version 6”, by C. Lee et al.
University of California/Berkeley
Internet – A Success
• Underlying technique
– IP – addressing and routing
– TCP/UDP – data transmission control (e.g.,
error recovery, flow control)
• Application
–
–
–
–
WWW (killer application)
E-mail
Telnet
Chat
University of California/Berkeley
Internet – A Success
• Underlying design
–
–
–
–
Connectionless datagram switching
Stateless end-to-end principle
Best effort
Client server model
• Less assumptions  more scalable & robust 
easy to develop
– Cost: some performance loss (e.g. transmit data over a
network whose MTU >> 576 bytes – maximum packet
size in IPv4)
University of California/Berkeley
Internet = Mail System
Int.
…
A mail system
USA
Bottleneck
Berkeley Oakland
Los Angeles
New York
…
University of California/Berkeley
Internet = Mail System
Int.
A better mail
system
…
USA
North CA
…
South CA
Los Angeles
Berkeley Oakland
…
…
…
University of California/Berkeley
Problems in IPv4
• Scalability
The most urgent thing!!
– Address run out
– Explosive routing tables (router is the bottle
neck of Internet, instead of network speed)
• QoS
– Best effort is not enough
– Commercialized Internet
• Security
University of California/Berkeley
Address Run Out
• 232 = 4,294,967,296, will run out before 2005
1000000
900000
800000
700000
600000
500000
460 Million users
400000
300000
200000
hosts
100000
0
2005
2003
2001
1999
1997
1995
1993
1991
1989
Source: Cerf, based on www.nw.com, Jan 2000
• “32 bits should be enough address space
for Internet” – Vint Cerf, 1977
– 32 bit address space is approximately 107 times
of the # of computers in DARPA time.
University of California/Berkeley
# Of Items In A BGP Routing Table
Projected routing table
growth without
CIDR
Moore’s Law and CIDR
made it work for a while
Deployment
Period of CIDR
University of California/Berkeley
Effort On Saving IPv4
• VLSM(Variable Length Subnet Mask)
– Try to figure out “problem of triple bears”
•
•
•
•
CIDR(Classless Inter-Domain Routing)
NAT(Net Address Translation)
L3 Switching，MPLS
RSVP、RTP/RTCP、DirectRoute、SSL
• However, due to scalability reason, a new IP
protocol has to be developed
University of California/Berkeley
What Do IPv6 Do?
• Address
– 128 bits. How large it is?
• ~ 3×1038
• Suppose earth as a smooth sphere, then there are one mol
(6.02×1023) IPs/m2
– Why 128 bits?
– Unicast, multicast, anycast
– For one interface, it can have multiple IPv6 addresses
• Routing
– Prefix routing and aggregation (based on CIDR)
– Address space is strictly aggregated
– Fixed size based header
University of California/Berkeley
Difference In Header
University of California/Berkeley
What Do IPv6 Do?
• MTU: 576 bytes  1280 bytes
• Type of Class (8 bits) and Flow label (20
bits) fields in header
• Mobile IP
– Redirect the route to the mobile node if
needed
• Security architecture
– Protection for key header
University of California/Berkeley
What Do IPv6 Do?
• Network management
– Neighbor discovery
• MTU
• Address resolution
• Network prefix
• Address lifetimes
– Address autoconfiguration
• Use 64-bit IEEE EUI-64 address of the hardware
• Network prefix + 64-bit hardware address
University of California/Berkeley
IPv4  IPv6
• Won’t happen in one day
• Dual protocol stacks
• Currently, 6bone uses IPv6 over IPv4
tunnel to connect IPv6 nodes
IPv6 node
IPv6 node
IPv4 world
University of California/Berkeley
Discussions
• IPv6 changes the underlying technique of
Internet, then what will be the change in
application? What will be the killer
application in future?
• In past, we have IPv4, then apps comes
out; how about today’s situation?
University of California/Berkeley
Summary
• Internet is a success
• IPv4 has problems, especially in address
space, routing, QoS and security
• IPv6 want to address those problems
• It may be a long time for IPv4 migrating to
IPv6