Internet Protocol Version 6

Internet Protocol Version 6
Internet Protocol Version 6
Arne Schipper
[email protected]
Reykjavı́k University
ABSTRACT
This paper describes IPv6, the new Version of the Internet
protocol. Currently IPv4 is the standard for the Internet,
there are just some few research projects and encapsulated
nets that make use of the new version of the Internet Protocol. The main differences and advantages in comparison
to IPv4 will be addressed, clarified with examples. Some
outlook will be given, considering when and maybe even if
the new version will be remarkable in use.
Keywords
IPv4, IPv6, Internet Protocol, IPng
1.
INTRODUCTION
The Internet (abbr. Interconnected Networks) is a worldwide network of independent networks. The main purpose
is to provide facilities for communication and information
exchange. Every computer of the Internet can basically communicate with every other. This is made available by defined
protocols. Very often people use the world wide web (www)
synonymic for the Internet, though this is just a part of it.
The Internet emerged from the ARPANET, with was founded by the Advanced Research Project Agency of the US
Department of Defence in 1969. This research agency made
first attempts with packet switched networks, to reduce the
error-proneness and improve the workload of normal circuit
switched networks. It took long time to establish the Internet as the institution as we know today. Many different protocols, institutions and applications were created and used.
First, the network was used to interconnect American universities researching for military purposes. In 1983, when
much non military information flew through the ARPANET, the ARPANET became separated into a scientific
ARPANET and a military MILNET, connected by gateways. One year earlier, the decision was made to switch the
ARPANET to TCP/IP, to allow easy connections to the
CSNET, another research network. In 1990 the ARPANET
was switched off, the net access moved more and more to
commercial providers and universities were given funds to
receive net access.
Since 1983 the Internet Protocol suite became the only protocol used for the Internet. Many different, sometime proprietary solutions tried to gain distribution, but alltogether
failed. So today, the Internet Protocol (IPv4) is, apart from
IPv6, the only standard used throughout the Internet.
In this paper there will be first provided a explanation of the
currently used version, the IPv4. The general architecture
will be covered and it will be considered inside the Internet
protocol stack. Then the main drawbacks of the IPv4 will
be listed and discussed. Another chapter will explain, how
with these drawback was coped to maintain the working of
the Internet. Different extensions for IPv4 will be covered.
The main chapter will introduce IPv6, its specification and
how all the drawbacks of the previous version were faced. In
the last chapter some outlook will be given how and when
the Internet Protocol version 6 will be introduced.
2. INTERNET PROTOCOL VERSION 4
2.1 The IP in the Internet protocol stack
The Internet Protocol is just one of the protocols providing
the functionality of the Internet. All the protocols used to
operate the Internet can be described in a protocol stack,
which consists of different layers for each sort of service. This
structure is similar from the ISO/OSI reference model, but
the actual implementation is slightly different. The Internet
mainly consists of four layers, as described in figure 1
Figure 1: Internet Protocol stack [37]
The lowest layer (we do not look an the physical layer like
wire, glass fibre or Carrier pigeon [38]) is called data link
layer. Mostly Ethernet is used for the implementation of
that layer, but it is also possible to utilise Token ring, ISDN,
802.11 WiFi, PPP, etc. The main purpose of that layer is to
pass packets from the network layer to the desired destination host within the same network. No routing is provided
by this layer, neither are any guaranties given that the data
will reach its destination. The process can be controlled in
a network interface card firmware, or by software drivers.
The next layer, called network layer, makes use of the Internet Protocol, currently mostly IPv4. The main task of
this layer and the Internet Protocol is to move so called IPpackets from a source to a destination. Addressing is done
using IP-addresses. The main benefit of that protocol is
that it can operate over the physical borders of a single network using routers. The Internet Protocol is that protocol
which defines the Internet as a group of loosely connected
computers, as it defines a end-to-end protocol. This protocol is rather simple, that is also a reason why it became
such widespread, but another reason for lacking of certain
services.
These services, like guarantied in order delivery, are provided by the next layer, the transport layer. A protocol
providing these services is the Transport Control Protocol
(TCP), which is in wide use in the Internet, resulting in the
term TCP/IP protocol suite, which makes up the Internet.
Another famous protocol on that layer is User Datagram
Protocol (UDP), which is much easier and faster, as it does
not guaranty delivery like TCP.
Using all these techniques, the application layer is on top of
the stack. Here are the famous protocols like HTTP, SMTP,
SSH, BitTorrent, telnet, which made the Internet to what it
is now.
2.2
2.2.1
IPv4 in detail
History
When the Internet Protocol was designed, the circumstances
regarding the number of hosts and the net topology were
quite different from nowadays. In the very beginning in
1969, just 4 computers were connected to each other. This
number grew only moderately during the next few years [33],
which can be seen in figure2. So designing a protocol in that
times was quite different from what would have happened
today, or even a decade ago. A main issue then was to
reduce the amount of memory space needed. When designing IPv4 [17] the authors thought that 232 addresses should
be sufficient. Some other features implemented in this first
version were never or hardly used, whereas some other important features were missing and had to be incorporated
into the existing structures.
2.2.2
Figure 2: Number of hosts connected [4]
Figure 3: Formats of classed IP addressing [26]
Thus resulted in a relatively small number of so called class
A subnets (namely 124), but each net could contain ≈ 16.7
M hosts. In the 16,384 class B nets it is possible to have
65,534 hosts, whereas in the 2,097,152 class C nets it is possibly to work with 254 hosts. Due to this architecture, there
were problems in assigning new IP addresses to new hosts,
but this will be discussed further in Chapter 2.3.1.
2.2.4
Address
127.0.0.0/8
10.0.0.0/8
172.16.0.0/12
192.168.0.0/16
224.0.0.0/4
255.255.255.255
Address format
IPv4 uses 32 bit addresses, which leads to a theoretical possible number of 232 ≈ 4.3 Mia addresses. Addresses are
normally displayed in dot-decimal notation, that means in
4 decimal numbers of 8 bits, e.g. 199.222.69.111. Other
notations are hexadecimal and octal representations.
2.2.3
Address allocation
Originally, the IP address was divided into two parts: A
network part and a host part.
Special address ranges
There are several address ranges reserved for special purposes. These reservations were not there from the beginning of the Internet Protocol. The most popular address
ranges are listed below in table 1, note that they were not
reserved from the beginning, but were slowly added to the
IPv4. They were summed up in several RFC, as for example
in RFC 1700 [27].
Use
loopback device
private class A network
private class B network
private class C network
multicasts
broadcast
Table 1: Some IP address ranges
2.2.5
IP packet format
IP packets consist of a header and the payload data. The
actual interesting of an IP packet is the header, see figure
4. It contains several fields with protocol information. The
most important fields are the source and destination address,
both 32 bits fields, as explained above. The other fields hold
information about the version (4 for IPv4), whole length of
the header, type of service, checksums and more information
like TTL, time-to-live. The type of service (TOS) field was
intended to be used for precedence handling of certain IP
packets, but that was very seldom implemented. The field
total length indicates the total length of the whole IP packet,
that means header and payload data. As this is a 16 bit
field, the maximun length of an IP packet is limited to 64
kB. This is hardly used, as most communication goes over
ethernet, and and fragmenting is slow, the size of an IP
packet is adjusted to the payload size of an ethernet frame,
often called MTU (maximum transfer unit), which is 1500
bytes. It is possible, to use extra option fields of the IP
header, each of a size of 32 bit, this must be announced in
the IHL field.
Figure 5: Number of hosts in the DNS [4]
With the initial design of the Internet Protocol, namely the
differentiation of the address in a host a network part, it
was quite easy for the big Internet routers to route the IP
packets to the desired networks. But with the changes made
to the protocol, see chapter 2.4, another problem arose. The
main routing tables were growing bigger and bigger. The
router needed to hold the information about the next router
for every destination address in memory, because of speed
concerns. With the growth of the routing tables that became
a remarkable problem. With almost every new host in the
Internet, new entries had to been make. This resulted in
tables sizes as shown in figure 6. In the tentative draft for the
routing tables the designer chose a table size of 10.000 [16],
which was to be exceeded in January 1993 and a big problem
had top be solved.
Figure 4: IPv4 header [34]
Directly after the header begins the payload section. Depending on the header size, and assuming a MTU of 1500
bytes, we can use up to 1480 bytes for data, that is a ratio
of 98.6 %. If a TCP datagram is sent, the the TCP header
and the TCP payload data will be inside this IP payload
frame. This actually leads to a lower ratio of payload data.
2.3
Drawbacks of IPv4
During the short explanation of IPv4 some drawback were
already mentioned. To understand why a new Internet Protocol is necessary, it is importend to draw out more disadvantages and to discuss them.
2.3.1
IP address depletion
As mentioned before, it is theoretically possibly to assign ≈
4.3 milliard IP addresses to host. Due to very unrestricted
allocation of IP addresses, that is for example be assigning a complete class A network with 16.7 million addresses,
there were nor many addresses left. There were different
predictions of running out of class B addresses around the
year 1994 [16]. As big parts of the subnets were not used,
estimations were predicting only 240 million useable IP addresses [12], though this should be higher today. Nevertheless this number is today be far exceeded, as can be seen in
figure 5.
2.3.2
Growing routing tables
Figure 6: IPv4 Internet route table entry trend [1]
2.3.3
Complicated header format
The header was initially designed to be flexible and extendable. There the decision was made to allow additional
header fields, as described above. With other fields like
header length and checksum, there was little room for efficient (hardware) routing, as the header had to be examined
exactly: it was necessary to analyse the optional fields and
to build the checksums.
2.3.4
Security problems
With IPv4 it was not possible to establish secure connections. Securing the channel had to be done by extensions or
the upper layers, like IPSec or SSH. Therefore the communication partner you want to talk to also needed the same
protocols, a condition which was not always met.
2.3.5
Bandwith reservation problems
Though there is a field in the IPv4 header (ToS) which
should be used to specify the priority of a package, it was
hardly used. Some research groups suggest to use this field
with techniques like Differentiated Services [19], but this is
not quite sophisticated yet and just a workaround applied
to the given structure of IPv4.
2.3.6
No multicast
The initial design of IPv4 did not make any arrangements
for multicast. That became a problem when huge amounts
of data had to be transmitted to a group of hosts, e.g. video
streaming. Rather than using one single-to-many connection, several single-to-single connections had to be used,
which stressed the network.
2.3.7
Bad mobile support
With the growing numbers of mobile applications, and the
demanding from them for the Internet Protocol, a new problem arose. The roaming of connected mobile devices is still
not easy. For IPv4 there is one technique, called Mobile
IP [36, 23], but that works with so called home mobile agents
and has a certain overhead.
2.3.8
Renumbering
Renumbering, that means if you want to change you IP address, produces today a lot of overhead. Changing the address sometimes meant to be multihomed for a short time,
in order not to loose connectivity to your customers. That
also had influence on the size of the routing tables and the
performance as well.
2.4
IPv4 now
Would all these problems have remained until now, the Internet just would not work anymore. To cure all these drawbacks, many extensions to the protocol were made. They are
described in a not negligible number of RFCs and resulted
in a rather blown up and patched protocol.
The problem with the depletion of IP address was solved by
stricter allocation of addresses. It were not just complete
subnets like a class B net which was assigned, but smaller
parts. Therefore the subnet mask was established and the
old system with the network and host part was abolished,
see CIDR [32]. It is now possible to run subnets of almost
arbitrary size, resulting in a much better usage of addresses
in a certain subnet. Another positive contribution to the
address problem was the strong usage of Network Address
Translation (NAT), which hided a whole (company) subnet
behind just one IP address. That was on the one hand done
to save IP address, but also to gain security of the data
coming to and leaving from the subnet, as there was just
one point were the data had to be controlled. One should
mention that that caused other problems, like a single point
of failure, if the NAT router crashed or problems if you want
to be reachable from the outer net. And the general idea of
the Internet, the end-to-end connection of the independent
hosts was violated with this method.
The router table problem was faced by using a new technique called route aggregation, introduced with BGP-4 [25].
Pathes to subnets were aggregated in just one routing table
entry. Roughly spoken that means, if there were before 3
entries in a table (lets say 197.8.0.23, 197.8.1.25, 197.8.2.27)
and we knew that they belong to the same net, we just would
make one entry out of them, namely 192.8.0/22, indicating
that all packets belonging to that subnet should be routed
to the same router. That required also a change in the allocation of addresses, meaning that coherent address ranges
had to be given to the same provider.
Multicast was implemented later, but remained withheld
to local net structures. Some multicast networks were implemented, like the MBone [39], but that was just a bad
workaround for the real problem.
Several other drawbacks were also faced, as already mentioned above. Sometimes the developers just used the layers
above to implement functions and features that IPv4 was
lacking in their opinion, like IPSec or different Quality of
Service (QoS) functions. Mobile IP is a nice extension, but
just a add on and would scale well. As the header could not
be changed, the inefficient layout is still used.
3. INTERNET PROTOCOL VERSION 6
3.1 The birth of IPv6
In the beginning of the 90s many people realised, that it was
necessary to develop a new protocol for the Internet, due
to the limitations of the very old and for these times not
appropriate IPv4. An IPng (IP next generation) effort was
started to solve these issues. After long discussions, which
can be found in several RFCs, starting with RFC 1550 [5],
IPv6 [8] was elected to be the new protocol in 1995, as it was
the merger of many good ideas. But the struggles did not
stop, resulting to some redesigning of that protocol, leading
to the final specification in RFC 2460 [10]. These discussions
and changing of the protocol was one main drawback, as
already finished implementations had to be condemned.
3.2
What is new
In the design phase of IPv6 the mail goal was to solve all the
problems that arose during the usage of IPv4 in the growing
Internet and provide future-proofness. So the main changes
can be directly adapted from the downsides of IPv4.
3.2.1
New addresses
The most remarkable change of course took place in the
header, as the header contributes most to the def intion
of a communication protocol. When designing the header
the persons in charge were aware of the insufficient address
length of version 4, so they extendend the address length
to 128 bit. That results in the unimaginable big number
of theoretical 3, 4 × 1038 addresses. It is possible to assign
6, 89×1017 to each square mm of the earth [35]. But that are
theoretical numbers, as the IP addresses are again separated
in different parts.
3.2.1.1
Address format
A close look to the format of the new addresses clarifies
that. To understand it, it is necessary to know how the
new addresses are represented. Notation is not any more
dotted-decimal, but in 8 hexadecimal blocks of 16 bit each,
separated by a colon. An example would be:
Unicast:
Anycast:
Multicast:
An identifier for a single interface. A packet sent
to a unicast address is delivered to the interface
identified by that address.
An identifier for a set of interfaces (typically belonging to different nodes). A packet sent to an
anycast address is delivered to one of the interfaces identified by that address (the ”nearest”
one, according to the routing protocols’ measure
of distance).
An identifier for a set of interfaces (typically belonging to different nodes). A packet sent to a
multicast address is delivered to all interfaces
identified by that address.
Table 2: IPv6 Address Type Identification [15]
fier (TLA), Reserved for future use (RES), Next-Level Aggregation Identifier (NLA ID), Site-Level Aggregation Identifier (SLA ID) and given in these blocks to international and
national registries or Internet service providers (Figure 9).
That should result in a more efficient routing, but would also
have created islands in the Internet with common prefixes.
IP addresses would have been geographically dependent.
Figure 7: IPv6 header [34]
2001:0db8:85a3:08d3:1319:8a2e:0370:7344
To simplify the notation when dealing with long addresses,
it is possible to leave out blocks with zeros and write instead
two colons: 2001:0db8::1428:57ab is equal to
2001:0db8:0000:0000:0000:0000:1428:57ab, but
2001::25de::cade is a wrong notation, as it is not clear how
many groups of zeros were left out where.
3.2.1.2
Address types
There are, like in IPv4, several address types defined in IPv6,
each with different semantics, especially to the routers. Figure 8 shows the general addressing architecture of IPv6,
when the type of address is not specified. That is just a
general scheme and has no practical meaning.
|
n bits
|
128-n bits
|
+---------------------------------------------+----------------+
|
subnet prefix
| interface ID
|
+---------------------------------------------+----------------+
Figure 8: General addressing architecture[15]
In IPv6 there exist three types of addresses, which is shown
in table 2:
New is the anycast address. Note that broadcast is handled
via multicast, by multicasting to the all-nodes group.
The most interesting addresses are of course the global addresses, as they have to be routed through the whole Internet, and during the design phase of IPv6 the problem of the
big routing tables was aware.
In earlier implementations the developers thought of a hierarchically structured Internet, where complete address
ranges would be separated in Top-Level Aggregation Identi-
| 3| 13 | 8 |
24
|
16
|
64 bits
|
+--+-----+---+--------+--------+-------------------------------+
|FP| TLA |RES| NLA
| SLA
|
Interface ID
|
| | ID |
| ID
| ID
|
|
+--+-----+---+--------+--------+-------------------------------+
Figure 9: Addressing architecture, July 1998[9]
Another proposal for the address architecture removed this
strong separation in 2003. As it is apparent from figure,
is it just distinguished between a global routing prefix, a
subnet ID, and the interface ID. Depending on the type of
addresses, the sizes of the fields vary, as indicated by n and m.
The designers stated that it is up to the service providers and
allocation authorities to allocate meaningful address spaces,
that allow route aggregation like the CIDR and BGP-4 in
IPv4.
|
n bits
|
m bits |
128-n-m bits
|
+------------------------+-----------+-------------------------+
| global routing prefix | subnet ID |
interface ID
|
+------------------------+-----------+-------------------------+
Figure 10: Addressing architecture, April 2003[14]
Table 3 gives a complete overview over the type of addresses. Global unicast addresses were already explained
above. Analogical to IPv4 there is the unspecified address,
when the interface is waiting for an address, and the loopback address, intended to send messages to itself. It is important that the semantics of IPv6 imply that we do not
assign an address to a host or node, but to an interface.
This becomes clearer when looking at the requirements a
host must meet before it can communicate over IPv6.
Anycast addresses are taken from the unicast address spaces
(of any scope) and are not syntactically distinguishable from
unicast addresses. A possible use could be to discover all the
routers attached to a particular subnet.
Multicast addresses have generally the same meaning like
in IPv4, it is possibly to reach a group of host by sending
Address type
Unspecified
Loopback
Multicast
Link-local unicast
Global unicast
Binary prefix
00...0 (128 bits)
00...1 (128 bits)
11111111
1111111010
(everything else)
IPv6 notation
::/128
::1/128
FF00::/8
FE80::/10
Table 3: Address Type Identification [15]
just one packet, which saves bandwidth and router load.
In IPv4 that was just possible in the local network, or via
workarounds like the MBone, which did not scale. In IPv6
it is possible to set options when sending to a multicast
group, so that the packets are routed through the Internet.
In the current RFC 4291 [15] there are currently six special
multicast scopes given:
• Interface-Local scope
• Link-Local scope
• Admin-Local scope
• Site-Local scope
• Organization-Local scope
• Global scope
These will be routed through the Internet. So it may be
possible to send a multicast request to all NTP time servers
that joined the global group in the whole Internet, by specifying a global scope multicast address.
Link-Local addresses are for use on a single link. They have
the 10 bit prefix as described above, then 54 bit of zeros,
and the last 64 bit contain the interface identifier. They
are used on single links and for automatic address configuration. Packets with these addresses must not be forwarded
by routers.
To identify himself, a host needs to meet certain criteria for
his network interface card. The card has to have at least:
3.2.1.3
3.2.1.4
• Any additional Unicast and Anycast addresses that
have been configured for the node’s interfaces (manually or automatically).
• The loopback address.
• The All-Nodes multicast addresses
• The Solicited-Node multicast address for each of its
unicast and anycast addresses.
• Multicast addresses of all other groups to which the
node belongs.
Interoperability with IPv4
There are special address formats reserved in the IPv6
header, that guarantee interoperability with the old Internet Protocol. These special addresses belong to the global
unicast address range. They start with a sequence of 80
zeros, and then 16 ones, and then the normal 32 bit global
IP address. These addresses are called IPv4-Mapped IPv6
addresses and provide one part of the interoperability issues
between IPv6 and IPv4.
3.2.2
Simplified header format
As obvious from figure 7, the header is much simpler and
does not contain as many fields as the header of version 4.
That will lead to a much faster treatment by the router,
resulting in more throughput. Many things were just left
out, like fragment information checksum information and
header length. The (main) header has now a defined length
of 320 bit, additional headers will be located between the
main header and the payload. This distinction make the
header on the one hand efficiently routable, but provides
enough flexibility for new extensions. IPv6 packets should
not be fragmented, the size of a single IPv6 packet must be
chosen by the sender according to the MTU of the data link
layer. Checksum information was removed, too, that issue
should be handled by layer 2 or layer 4 (e.g. TCP). All
these simplifications will increase the speed when a packet
has to cross many routers. The semantics of the old TTL
field has now changed to the actual meaning, HOPs, which
defines the maximum number of routers that can be passed.
The original meaning of TTL was the time a packet should
live. Flowlabel and class in the IPv6 header will be used for
Quality of Service, advising the routers how to prioritize the
packet.
3.2.3
• Its required Link-Local address for each interface.
Interface Identifiers
The 64 bit interface identifier is, in case we use ethernet,
created out of the 48 bit MAC address of the network interface card with a simple mapping and should be unique.
As this identifier can be seen in the public IP address, some
suggestions were made not to loose privacy with this mechanism ??. Now, the interface identifier is made with special
algorithms, ensuring obfuscation of the real hardware address. For other data links as the ethernet there are special
algorithms to generate this identifier.
Address autoconfiguration
It is still possible to configure a host manually, or to techniques similar to IPv4 DHCP. But new is that a host can
generate itself a valid IP address [28], by using his interface identifier and the prefixes announced by present routers.
Even if now routers are present, it is possible to communicate with neighbours without any configuration [30]. These
techniques can be even combined with DHCPv6 [11] to gain
additional information like nameservers.
3.2.4
Renumbering
Renumbering is made much easier. The details are quite
complicated and are nor scope of this document. Ressources
can be found in RFC 4076 [29].
3.2.5
Security features
In the first implementations of IPv4 there were no security
mechanisms that were assuring that a packet could not be
altered and seen during the way through the network. Now
there are extensions like IPSec, but they are not mandatory.
Both hosts have to implement the used security protocols. In
IPv6 a technique similar to IPSec is used, and is mandatory
for the IP protocol stack. This is made possible by one of
the extension headers. There is now real data security on
the network layer. Problems that also arose with IPSec and
IPv4 were the limitations due to the use of NAT, as the real
sending address was changed by the NAT router. NAT is
generally not necessary with IPv6 and should not be used.
Instead it is recommended to use other filter techniques to
protect certain networks.
3.2.6
Quality of Service
The Quality of Service capability of IPv4 was quite bad and
inefficiently solved. For upcoming applications that was not
sufficient. In IPv6 there are two fields reserved to provide
QoS functionality, though it is yet not completely clear, how
the exact implementation will be. For the flow label field
there is one proposal in RFC 3697 [18]. The other useable
field is traffic class.
3.2.7
Mobile IP
Mobile applications are growing, there are estimations for
over 2 milliard mobile users in the year 2007. To serve them
all with the services desired, it is necessary to create the
underlying infrastructure properly. Many see the IPv6 as
one of the backbone architectures for these growing market [21, 22, 13]. With the larger address space all mobile
devices imaginable could be addressed. But IPv6 provides
also good strategies when moving the devices from one network to another. There are some solutions for IPv4, like
home-agents and foreign-agents [23], which is also possible
for IPv6. There are new researches on how to use the underlying IPv6 architecture to simplify and enhance the home
and foreign-agent method by just sending some messages
over these relays and then switch to a direct connection,
making use of the IPv6 extension headers [7].
3.3
Advantage of IPv6 over IPv4
To sum up the advantages of IPv6 over IPv4 it is almost
sufficient to reread the previous chapter. When designing
the new version, all the drawbacks were in mind and so the
best attempts were many so solve them. Big advantages are,
as already pointed out, the great mobility support, the QoS
enhancement, the routing table aggregation and the larger
address space. Though many of these problems very later
solved in extensions to the IPv4 protocol, the old protocol
looks a a piece of software that was just patched for over 20
years. The IPv6 is much more elegant and redesigned almost
from scratch. As it will not be necessary to use NAT, it will
be much easier to use peer-to-peer applications like VoIP
and Instant Messaging - a definite future trend - as the endto-end paradigm is restored. Due to the auto configuration
mechanisms, the network administration costs of companies
will also be reduced.
3.4
3.4.1
Switching to IPv6
Situation
A disadvantage is obviously the migration or switching from
IPv4 to IPv6. There is an estimation about the costs of
about $75 milliard just for the United States. This should
not be considered as the biggest problem, as the infrastructure has to renewed in certain intervals nevertheless. Some
reports even state out that upgrading to the new version
would yield in just the market for durable goods (e.g. automobiles) to benefits of over $3 milliard each year [31].
Another hard issue is the fact that the current IPv4 with
all its extensions is working well for many companies and
that they do not see the need to switch. In North America
and Europe there are, due to historic reasons considering
the allocations of IPv4 address ranges, still enough IPv4 addresses left, so the need is much lower than for example in
Asia. This continent is a driving force for the IPv6 deployment, but many decisions are taken by Europe and the US.
The migration phase will also be hard to manage. It will
not be possible to switch the whole Internet from one day
to another from version 4 to version 6. So there will be a
migration phase, were IPv4 and IPv6 will coexist. There are
many methods to face these problems [24]. One explained
above is that the designers of IPv6 reserved some special
addresses in the IPv6 address range for IPv4-mapped IPv6
addresses. With these addresses it is possible for a host to
make a DNS query for the IPv6 address of an IPv4 host.
These methods can be used within dual stack hosts, which
support both IPv4 and IPv6. Other techniques are tunneling, where the IPv6 packets are packed into an IPv4 packet
in order to be able to use the existing IPv4 infrastructure.
The packets will be unpacked and processed at the destination host. It is also possible to use a kind of proxy at some
border points of networks, where the different versions are
translated.
To support both IPv4 and IPv6 at the same time, it is important to get the right addresses when performing a DNS
query. BIND [?], the most used DNS server software, supports both IPv4 and IPv6 queries. The IPv4 addresses are
stored in a so called A record. To support Ipv6 an AAAA
record was created, which holds the IPv6 address of a certain
host.
3.4.2
Which OS is IPv6 ready
The following list gives a small overview over common operating systems and their efforts towards IPv6:
• BSD - At the moment the best support for IPv6 [3],
but no IPv4-to-IPv6-Mapping implemented.
• AIX - Since version 5.2 even with Mobil IP
• Cisco - Productive support since IOS version 12.3
• Linux - similar support like BSD
• Mac OS X - Support since version 10.2
• Windows XP - must be explicitly turned on, but it is
possible since SP2
4.
CONCLUSION AND OUTLOOK
Today almost all traffic of the Internet is done with IPv4.
There are no other protocols, if we leave out the few IPv6 islands that are used in the Internet. A switching, respectively
migrating means that IPv4 would be replaced by IPv6. As
mentioned above, there are different groups of interests. To
many companies and other persons using the Internet changing just means no benefits, but high costs [31], as the have
so buy new soft and hardware. The actual version normally
gives them the opportunities they need.
On they other hand there are the big software, hardware
and mobile application manufactures. They clearly see the
need for a better protocol, facing problems like bad mobility support, lack of security features, address depletion in
certain areas of the world. They have s strong desire in deploying the IPv6 quite soon. It is very important for the
success of a new technology how the great companies with
the biggest market-share stand toward it. Major players
like Microsoft, Cisco and mobile telephone companies are
investing big sums in research for IPv6. With the support
of the new protocol in MS Windows a big entry barrier has
been removed, due to the spread of Windows on consumer
computers.
Governments and several organisations, as well as big companies, are preparing studies in order not to loose track
in the recent development. Countries in Asia, like Japan,
China and Korea are much further in using and testing IPv6.
One reason is the limited amount of IPv4 addresses they received. If the growth of the economy in China continues,
there will soon be 500 million mobile phone users and an
equal number of Internet users. That will not be possible
with the IPv4 addresses available.
The world can not close its eyes to this new technology.
Future trends are demanding special infrastructure like security and mobility. Devices will be always connected to
the Internet (always on), and will keep their address. Applications like VoIP, peer-to-peer and smartphones will gain
great success. That is simply not possible with IPv4.
In spite of the costs that we are facing, the question is not
anymore if and if yes, when, but when. And the date, when
the last IPv4 application will stop operating is coming nearer
and nearer, and always faster.
5.
REFERENCES
[1] AS1221 BGP Table Statistics.
http://bgp.potaroo.net/as1221/bgp-active.html.
[2] Join Uni Münster. http:
//www.join.uni-muenster.de/Join/index_join.php.
[3] KAME Project. http://www.kame.net/.
[4] www.isc.org.
http://www.isc.org/index.pl?/ops/ds/hosts.php.
[5] S. Bradner and A. Mankin. RFC 1550 - IP: Next
Generation (IPng) White Paper Solicitation.
http://www.ietf.org/rfc/rfc1550.txt, December
1993.
[6] Cisco Systems. The Coming Internet Evolution: IPv6
And Its Implications For The Service Provider
Marketplace. Technical report.
[7] D. Johnson, C. Perkins and J. Arkko. RFC 3775 Mobility Support in IPv6.
http://www.ietf.org/rfc/rfc3775.txt, June 2004.
[8] S. Deering and R. Hinden. RFC 1883 - Internet
Protocol, Version 6 (IPv6) Specification.
http://www.ietf.org/rfc/rfc1883.txt, December
1995.
[9] S. Deering and R. Hinden. RFC 2373 - IP Version 6
Addressing Architecture.
http://www.ietf.org/rfc/rfc2373.txt, July 1998.
[10] S. Deering and R. Hinden. RFC 2460 - Internet
Protocol, Version 6 (IPv6) Specification.
http://www.ietf.org/rfc/rfc2460.txt, December
1998.
[11] R. Droms. RFC 3736 - Stateless Dynamic Host
Configuration Protocol (DHCP) Service for IPv6.
http://www.ietf.org/rfc/rfc3736.txt, April 2004.
[12] A. Durand and C. Huitema. RFC 3194 - The
H-Density Ratio for Address Assignment Efficiency
An Update on the H ratio.
http://www.ietf.org/rfc/rfc3194.txt, November
2001.
[13] H. Einsiedler, K. Jonas, J. Jähnert, R. Schmitz and
M. Liebsch. Mobility Support for a Future
Communication Architecture. In IST Mobile
Communications Summit, pages 9–12, 2001.
[14] R. Hinden and S. Deering. RFC 3513 - Internet
Protocol Version 6 (IPv6) Addressing Architecture.
http://www.ietf.org/rfc/rfc3513.txt, April 2003.
[15] R. Hinden and S. Deering. RFC 4291 - IP Version 6
Addressing Architecture.
http://www.ietf.org/rfc/rfc4291.txt, February
2006.
[16] C. Huitema. Routing in the Internet. Upper Saddle
River, NJ: Prentice Hall, 2000.
[17] Information Sciences Institute. RFC 791 - Internet
Protocol. http://www.ietf.org/rfc/rfc791.txt,
September 1981.
[18] J. Rajahalme, A. Conta, B. Carpenter and S. Deering.
RFC 3697 - IPv6 Flow Label Specification.
http://www.ietf.org/rfc/rfc3697.txt, March 2004.
[19] K. Nichols, S. Blake, F. Baker and D. Black. RFC
2474 - Definition of the Differentiated Services Field
(DS Field) in the IPv4 and IPv6 Headers.
http://www.ietf.org/rfc/rfc2474.txt, December
1998.
[20] Microsoft Cooperation. Introduction to IP Version 6.
Microsoft Windows 2003 Server White Paper,
September 2003. Updated February 2006.
[21] Microsoft Cooperation. Understanding Mobile IPv6.
Technical report, April 2004. Updated November 2005.
[22] Moby Dick Consortium. Mobility and Differentiated
Services in a Future IP Network - Final Project
Report. Technical report, April 2004.
[23] C. Perkins. RFC 3344 - IP Mobility Support for IPv4.
http://www.ietf.org/rfc/rfc3344.txt, August
2002.
[24] R. Gilligan and E. Nordmark. RFC 2893 - Transition
Mechanisms for IPv6 Hosts and Routers.
http://www.ietf.org/rfc/rfc2893.txt, August
2000.
[25] Y. Rekhter and T. Li. RFC 1771 - A Border Gateway
Protocol 4 (BGP-4).
http://www.ietf.org/rfc/rfc1771.txt, March 1995.
[26] J. Reynolds and J. Postel. IP-Adressierung verstehen.
http://zeus.fh-brandenburg.de/~ihno/lehre/
internet/#IPADDR, 1994.
[27] J. Reynolds and J. Postel. RFC 1700 - Assigned
Numbers. http://www.ietf.org/rfc/rfc1700.txt,
October 1994.
[28] S. Thomson and T. Narten. RFC 2462 - IPv6
Stateless Address Autoconfiguration.
http://www.ietf.org/rfc/rfc2462.txt, December
1998.
[29] T. Chown, S. Venaas and A. Vijayabhaskar. RFC 4076
- Renumbering Requirements for Stateless Dynamic
Host Configuration Protocol for IPv6 (DHCPv6).
http://www.ietf.org/rfc/rfc4076.txt, May 2005.
[30] T. Narten, E. Nordmark and W. Simpson. RFC 2461 Neighbor Discovery for IP Version 6 (IPv6).
http://www.ietf.org/rfc/rfc2461.txt, December
1998.
[31] US Department of Commerce. Technical and economic
assessment of Internet Protocol Version 6 (IPv6).
Microsoft Windows 2003 Server White Paper, January
2006.
[32] V. Fuller, T. Li, J. Yu and K. Varadhan. RFC 1519 Classless Inter-Domain Routing (CIDR): an Address
Assignment and Aggregation Strategy.
http://www.ietf.org/rfc/rfc1519.txt, September
1993.
[33] Wikipedia, die freie Enzyklopädie. Chronologie des
Internets. http://de.wikipedia.org/wiki/
Chronologie_des_Internets, March 2006.
[34] Wikipedia, die freie Enzyklopädie. IP-Header.
http://de.wikipedia.org/wiki/IP-Header, March
2006.
[35] Wikipedia, die freie Enzyklopädie. IPv6.
http://de.wikipedia.org/wiki/IPv6, March 2006.
[36] Wikipedia, die freie Enzyklopädie. Mobile IP.
http://de.wikipedia.org/wiki/Mobile_IP, March
2006.
[37] Wikipedia, the free encyclopedia. Internet Protocol.
http://en.wikipedia.org/wiki/Internet_Protocol,
March 2006.
[38] Wikipedia, the free encyclopedia. IP over Avian
Carriers. http:
//en.wikipedia.org/wiki/IP_over_Avian_Carriers,
March 2006.
[39] Wikipedia, the free encyclopedia. MBone.
http://en.wikipedia.org/wiki/Mbone, March 2006.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement