ipv6book-chap1, EBOOKS matma, infa, chemia; tutoriale itp
[ Pobierz całość w formacie PDF ]
1
IPv6 Rationale and Features
Back in the 1970s, the Internet Protocol (IP) was designed upon certain assumptions and key
design decisions. After more than 25 years of deployment and usage, the resulting design
has been surprisingly appropriate to sustain the growth of the Internet that we have seen
and continue to see; not only the increase of the number of devices connected, but also of
the kinds of applications and usage we are inventing everyday. This sustainability is a very
impressive achievement of engineering excellence.
Despite the extraordinary sustainability of the current version (IPv4), however, it is suf-
fering and the Internet Protocol needs an important revision. This chapter describes why
we need a new version of the IP protocol (IPv6), by describing the Internet growth, the
use of techniques to temper the consequences of that growth and the trouble experienced in
deploying applications in current IPv4 networks. Some architecture considerations are then
discussed and new features needed in current and future networks presented.
Next, the work towards IPv6 at the IETF is shown along with the key features of IPv6.
Some milestones are also tabled. Finally, the IPv6 return on investment and drivers is
discussed.
1.1 Internet Growth
The origin of IPv6 work lay in the imminent exhaustion of address space and global routing
table growth; both could be summarized as Internet growth.
1.1.1 IPv4 Addressing
The Internet is a victim of his own success. No one in the 1970s could have predicted this
level of penetration into our lives.
Migrating to IPv6: A Practical Guide to Implementing IPv6 in Mobile and Fixed Networks
Marc Blanchet
© 2006 John Wiley & Sons, Ltd
2
Migrating to IPv6
log (number of objects using the network)/number of bits of the address space.
Based on some empirical studies of phone numbers and other addressing schemes, the author
concluded that this H ratio usually never reaches the value of 0.3, even with the most efficient
addressing schemes. An optimistic H ratio is 0.26 and a pessimistic one (for not very efficient
addressing schemes) is 0,14. At H
=
026, with an addressing of 32 bits, the maximum
number of objects, in the case of IPv4 the number of reachable hosts, is 200 000 000.
1
When
IPv4 Internet reaches 200 million reachable nodes, the IPv4 addresses will be exhausted.
Moreover, the IPv4 address space was designed with three classes (A, B and C)
2
which
makes the address space usage even less efficient than with the optimistic H ratio. In August
1990 at Vancouver IETF, a study [Solensky, 1990] demonstrated the exhaustion of class B
address space by March 1994. Figure 1.1 shows the summary slide presented during that
IETF. This was an important wakeup call for the whole Internet engineering community.
=
Figure 1.1
Solensky slide on IPv4 address depletion dates
1
RFC1715 was also used as input to define the IPv6 address length to 128 bits.
2
D and E classes also exist but are not for unicast generic use.
In theory, 32 bits of IPv4 address space enables 4 billion hosts. Studies [RFC1715] have
shown that the effectiveness of an address space is far less. For example, RFC1715 defines a
H ratio as: H
IPv6 Rationale and Features
3
At that time, most organizations requesting an address space pretty easily obtained a class
B address block, since there was plenty of IPv4 address space. Assigning class C address
blocks to organizations was the first cure; it decreased the initial address consumption
problem but introduced more routes in the global routing table, therefore creating another
problem.
1.1.2 IPv4 Address Space Utilization
Let’s talk about the current IPv4 address space utilization. The IPv4 address space is 32 bits
wide. IANA allocates by 1/256th (0.4%) chunks to regional registries, which corresponds
to a /8 prefix length or to the leftmost number in an IPv4 address. Since the 224.X.X.X
to 239.X.X.X range is reserved for multicast addressing and the 240.X.X.X to 254.X.X.X
range is the experimental class E addressing, the total unicast available address space is of
223 /8 prefixes.
Figure 1.2 shows the cumulative number of /8 prefixes allocated since the beginning of
IPv4. At the end of 2004, there are 160 /8 prefixes allocated, representing 71% of the total
unicast available address space.
In 2003, 5 /8 prefixes were allocated by IANA to the regional registries. In 2004, 9 /8
prefixes were allocated (80% annual increase). In January 2005 alone, 3 /8 prefixes were
allocated. If every year after 2004, we are flattening the annual consumption to the 2004
number (9 /8 prefixes: i.e. 0% annual increase for the next 7 years), then Figure 1.3 shows
the exhaustion of IPv4 address space (223 /8 prefixes) by 2011.
240
220
200
180
160
140
120
100
80
60
40
20
0
X
XX
Year
Figure 1.2
IPv4 cumulative allocated address space as of 2004–12
X
4
Migrating to IPv6
240
220
200
180
160
140
120
100
80
60
40
20
0
8
1
9
1
9
2
9
3
9
4
9
5
9
6
9
7
99
90
00
1
0
2
0
3
0
4
0
5
0
6
0
7
0
8
0
9
1
0
1
1
Year
Figure 1.3
Prediction of IPv4 allocated address space with flat annual consumption
If we are slightly more aggressive by increasing the annual consumption by 2 additional /8
prefixes every year after 2004, which results in an annual increase of 22%, then Figure 1.4
shows the exhaustion of IPv4 address space by 2009.
A 20% annual increase is pretty conservative, given that:
•
large populations in China, India, Indonesia and Africa are not yet connected;
•
world population net annual growth is 77 million people [Charnie, 2004];
•
all kinds of electronic devices are increasingly being connected and always on;
•
broadband connections incur permanent use of addresses instead of temporary addresses
when dialing up;
•
each 3G cell phone consumes at least one IP address.
On the other hand, mitigating factors may delay this exhaustion:
•
some class A are assigned but not used and therefore could be reclaimed;
•
as in economics, the rarer something is, the more difficult it is to get and more it costs,
slowing the exhaustion but instead creating an address exchange market.
Despite this, the IPv4 address shortage is already happening, and severely, because
•
organizations usually get just a few addresses (typically 4) for their whole network, limiting
the possibilities of deploying servers and applications;
•
some broadband providers are giving private address spaces to their subscribers, which
means the subscriber computers cannot be reached from the Internet.
8
IPv6 Rationale and Features
5
240
220
200
180
160
140
120
100
80
60
40
20
0
8
1
9
1
9
2
3
9
4
9
5
6
9
7
9 9
0
1
2
3
4
5
6
7
8
0
9
Year
Figure 1.4
Prediction of IPv4 allocated address space with incremented annual consumption
1.1.3 Network Address Translation
The most important change regarding IP addressing is the massive use of Network Address
Translation (NAT). The NAT functionality is usually implemented within the edge device
of a network, combined with firewalling. For example, most organization networks have a
firewall with NAT at the edge of their network and most home networks have a home router
which implements firewalling and NAT.
NAT maps multiple internal private IP addresses to a single external IP address.
3
By
allocating new external port numbers for each connection, essentially this NAT mapping
process extends the address space by adding 16 bits of the port address space.
Figure 1.5 shows a basic network diagram of a private network with 2 computers
(N1 and N2) and a public network, such as the Internet with one server (S). The private
network uses private address space [RFC1918]. When internal nodes N1 and N2 connect to
server S, the source addresses (10.0.0.3, 10.0.0.4) of the packets are translated to the NAT
external IP address (192.0.2.2) when the packet is traversing the NAT. Server S receives
connections coming from the same single source address (192.0.2.2), as if it comes from
one single computer.
Table 1.1 shows how the detailed process works based on Figure 1.5. When the packet
traverses the NAT, the source IP address and port are translated to the external IP address
of the NAT and a new allocated port, respectively. For example, N1 source IP address
10.0.0.3 is translated to 192.0.2.2 and the source port 11111 is translated to the new allocated
3
NAT can map multiple internal addresses to more than one external address, but for simplication we are discussing
the most current used case: multiple internal to a single external address.
9 0
[ Pobierz całość w formacie PDF ]