IP Futures

1
IP Futures

• There are problems with IP which are a result of
  the phenomenal growth of the Internet over the
  past few years
• as of 1994, over half of the class B addresses
  have been allocated
• 32-bit IP addresses are inadequate
• the current routing structure is basically flat,
  making routing tables too large
• Scalability is the problem!!

2
Classless Addressing

• The IP addressing space has been running out
• Subnet addressing (early 1980s) helped to
  conserve the IP address space
• Unnumbered networks and transparent routers
  followed
• In 1993 work began on developing a new version of
  IP
• In the meantime something needs to be done
• Classless addressing, supernet addressing, or
  supernetting is the technique currently being used

3
The Basic Idea

• Classless addressing take a complementary
  approach to subnet addressing
• Instead of a single IP network prefix for an
  organization, addresses assigned to a single
  organization are allowed to span multiple classed
  prefixes

4
Why?

• The IP addressing scheme does not divide network
  addresses evenly
• 17,000 class B addresses and 2 million class C
  address
• The Goldilocks problem caused most organization
  to request class B addresses

5
PTT Inc.

• PTT Inc needs to register to get an IP address
• Class C is clearly too small, plus I want to
  subnet, so I want a class B
• Instead of a single class B address, I get a
  contiguous block of class C addresses

6
ISPs

• Classless addressing was also meant to be used in
  a broader context
• The ISP assigns numbers to its clients

7
Routing

• Clearly classless addressing affects the way that
  routing is done
• At first glance it appears that routing tables
  will need to grow
• CIDR takes care of this
• RFC 1519
• Contiguous class C addresses are represented by a
  single entry
• ( network address, count)

8
Routing Tables

• So the entry
• ( 192.5.48.0, 3 )
• Specifies the network addresses 192.5.48.0,
  192.5.49.0, and 192.5.50.0
• If ISPs make up the core of the internet
• Routing tables become much smaller
• An ISPs routing table has entries for all of its
  customers, but only one entry for any other ISP

9
Class C Allocation Rules
Start End Location
194.0.0.0 195.255.255.255 Europe
198.0.0.0 199.255.255.255 North America
200.0.0.0 201.255.255.255 Central and South America
202.0.0.0 203.255.255.255 Asia and the Pacific
10
CIDR In Practice

• CIDR uses a bit mask to identify the size of the
  block
• For a block of 2048 addresses starting at
  128.211.168.0
• The mask would be FF FF F8 00
• New shorthand
• 128.211.168.0/21
• 255.0.0.0/8
• 255.255.0.0/16
• 255.255.255.0/24

11
New IP Versions

• Four proposals have been made for a new version
  of IP
• SIP, the Simple Internet Protocol. Proposes a
  minimal set of changes to IP that uses 64-bit
  addresses and a different header format
• PIP, larger, variable length, hierarchical
  addresses with a different header format
• TUBA (RFC1347), TCP and UDP with bigger addresses
• TP/IX (RFC1475), 64-bit addresses, changes TCP/UDP

12
References

• The May 1993 issue of IEEE Network (volume 7,
  number 3) contains overviews of the first three
  proposals, along with an article on CIDR.
• RFC1454 also compares the first three proposals

13
What is IPv6?

• IPv6, also called IPng (next generation) is the
  new version of Internet Protocol
• Currently we are using IPv4, which IPv6 was
  designed to be a successor to
• Designed not to take a radical step away from
  IPv4, but improve upon it

14
Why IPv6?

• IPv4 has been designed early in the 70s
• Many things have been added
• MobileIP
• QoS
• Security (IPsec)
• Others
• These were not designed in IPv4 from the start.

15
Design Criteria

• Number of addresses
• Efficiency in routers low and very high bandwidth
  (100G/ bytes)
• Security
• Mobility
• Automatic configuration
• Seamless transition
• No need to change hardware

16
Features

• Expanded addressing capabilities
• Header format simplification
• Increased support for modular options
• Multicast routing capabilities
• Security capabilities
• Expanded QoS capabilities

17
Address Space

• IPv4
• 32 bit addresses
• addresses assigned to nodes
• little unassigned address space
• relatively small address space

• IPv6
• 128 bit addresses
• addresses assigned to interfaces
• approximately 82 of address space unassigned
• huge address space

18
How Huge?

• Every human on the planet has enough addresses to
  create a network the size of the current internet
• Or
• Earths surface is about 5.1 x 108 square
  kilometers
• This means there are about 1024 addresses per
  square meter of the Earths surface
• Or
• If you assign addresses at the rate of one
  million every microsecond, it would take more
  than 1020 years to exhaust the addressing space.

19
Or

• Imagine Bill Gates fortune is 85 billion
  (8.5x1010) Take 1 trillion Bill Gateses
• Convert their fortune to pennies
• Assign 1x1012 addresses to each penny
• Takes 8.5x1036 addresses
• Youve just assigned 2.5 of the entire IPv6
  address space

20
Address Notation

• 128 bit address are too big to write in the
  dotted-decimal-octet format (i.e. 129.21.3.103)
• New notation is hexadecimal digits separated by
  colons every 16 bits.
• 5f1bdf00ce3ee200002008002078e3e3
• Can append decimal octet for easier IPv4
  mapping/compatibility
• FFFF206.62.226.33
• Can abbreviate filler 0 bits with

21
IPv6 Address Types

• Three types of addresses
• Unicast
• Destination is a single computer
• Multicast
• The destination is a set of computers, the
  datagram is delivered to each member of the group
• Anycast

22
Anycast

• An anycast address is one that is assigned to
  more than one interface (typically belonging to
  different nodes)"
• A packet sent to an anycast address is routed to
  the "nearest" interface having that address,
  according to the routing protocol
• Allocated from the unicast address space
• Not to be used as the source address
• Assigned to routers only - not hosts

23
Anycast Example

• Your ISP has two upstream connections with two
  different service providers
• Service provider X uses anycast address A to
  identify all of its routers
• If you want your packets to go through X, you can
  do a source route to A
• This will go to the "nearest" router in A, so
  even if the network topology changes, your
  packets will still go through X

24
Important Addresses

• Localhost - 127.0.0.1 - 1
• The unspecified address - 0.0.0.0 - or 00
• Multicast
• all nodes on link ff021
• all routers on link ff022
• all hosts on link ff023

25
Format of Global Unicast Address

• Format prefix (001)
• TLA ID (top-level aggregation identifier)
• NLA ID (next-level aggregation identifier)
• SLA ID (site-level agg. id. - e.g. Subnet ID)
• interface identifier - based on MAC address

26
Format of Multicast Addresses

• Flags is a set of 4 flags.
• The high order 3 flags are reserved.
• The low order flag indicates whether the address
  is a permanently assigned or transient multicast
  address.
• Scope indicates multicast scope e.g.
• 1 node-local
• 2 link-local
• 5 site-local

27
IPv6 Packet Format
Optional
Base Header
Ext Header 1
Ext Header N

DATA!!
40 Octets
28
IPv6 Packet Header
Vers
Traffic Type
Flow Label
Payload Length
Next Header
Hop Limit
Source Address
Destination Address
29
IPv6 Packet Header

• Followed by optional header extensions and data
  payload
• Addresses are 64-bit aligned - for speed
• Note the lack of a header checksum
• For speed

30
Traffic Class

• 8-bit field
• A way to identify and distinguish between
  different classes or priorities of packets
• Used to make more intelligent routing decisions
• Can be set or changed by forwarding routers
• Currently experimental

31
Flow Label

• 20-bit field
• A flow is a sequence of packets
• Distinguished by
• Source
• Destination
• Flow Number
• Used by a source to request special handling by
  routers for all packets in a flow

32
Payload Length

• 16-bit unsigned integer
• Length of the rest of the packet in bytes
• Includes header extensions and data block

33
Next Header

• 8-bit selector
• For header extensions
• If header extensions are present, this field
  indicates the type of the first one following the
  current header

34
Hop Limit

• Limit on the number of times a packet can be
  forwarded on the network
• Set by source
• Decremented by routers when packet is forwarded
• If zero after decrement, packet is dropped,
  ICMPv6 Echo Timeout sent to source
• Analagous to TTL in IPv4

35
Extension Headers

• Hop-by-Hop options
• Source Routing
• Fragmentation
• Authentication
• Encapsulating Security Payload
• Destination options
• Upper-layer headers

• Optional Components
• Used to provide more information to routers or
  destination
• Seven types of extensions defined

36
Options Headers

• Two types
• Hop-by-hop examined by every router
• Destination examined by the receiving node only
• Contains further information to be known to the
  appropriate machines.
• Format specified, but options themselves
  currently unspecified (as far as I can tell)

37
Fragmentation

• Fragmentation is an option header as it is the
  exception, not the rule
• Unlike IPv4, fragmentation must happen at the
  source, not at intermediate routers - for speed
• This means source must have the ability to
  perform Path MTU Discovery
• IPv6 requires that every link on the internet
  have an MTU of 1280 bytes or greater

38
Upper-Layer Checksums

• The transport layer in IPv6 places a
  pseudo-header into the header extensions.
• This includes source and (ultimate) destination
  addresses.
• Upper-layer checksums must include this
  pseudo-header in their calculation so the
  destination can validate these fields.
• TCP, UDP and ICMPv6 all use these pseudo-headers

39
Transition

• Allows for incremental updates
• e.,g. one subnet can upgrade at a time
• Easy addressing for IPv4 compatibility/mapping
• Encapsulation of packets
• IPv6 in IPv4 (i.e. 6Bone)
• IPv4 in IPv6
• Can have both IPv4 and IPv6 on same network
• Will take years to complete

40
IPv4 Compatibility

• Transition Mechanism includes a technique for
  dynamically tunneling IPv6 over IPv4
  infrastructure.
• IPv6 nodes that utilize this technique are
  assigned special IPv6 unicast addresses that
  carry an IPv4 address in the low-order 32 bits.
• This is an "IPv4-Compatible IPv6 address"
• ltIPv4 addressgt
• e.g. 129.12.3.103

41
IPv4 Mapping

• Used to represent the address of IPv4-only nodes
  (those that do not support IPv6) as IPv6
  addresses
• Called an "IPv4-mapped IPv6 address"
• FFFFltIPv4 addressgt
• e.g FFFF129.21.3.103

42
Routing

• Nearly identical to IPv4 routing
• Fixes class A/B/C routing problems
• All of IPv4's routing algorithms will work with
  minor extensions
• Some new functionality
• Provider selection
• Host mobility (route to current location)
• Auto-readdressing (route to new address)

43
References

• RFCs
• 2460 Internet Protocol, Version 6 Specification
• 2373 IPv6 Addressing Architecture
• 1933 Transition Mechanisms for IPv6 Hosts and
  Routers
• www.ipv6.org
• www.6bone.net