Study on IPV6 Migration - Summary / October 2002
Study from IDATE and Pardieu Brocas Maffei & Leygonie's Office
for L'Autorité de Régulation des Télécommunications
En téléchargement (
IPv4, the version of IP currently used on Internet,
was developed about 20 years ago. Its major weakness is its addressing
space, since, in IPv4, an address is defined on just 32 bits. The rapid
success of Internet and the acceleration of the consumption of IP addresses,
has led some to fear a shortage of IP addresses in coming years. For now,
the limits of IPv4’s addressing system have been pushed back using processes
such as address translation (NAT) and the CIDR routing diagram to aggregate
IP addresses.
IPv6, developed by the IETF in the mid Nineties, is the next version
of the Internet protocol. IPv6 improves on the addressing capacities of
IPv4 by allocating 128 bits to IP addresses instead of 32, thereby opening
up an almost infinite pool of IP addresses.
A number of events in recent months have led us to believe we will be
seeing a migration to IPv6 very soon: a number of government authorities
have stated their positions, in particular Asia, the European Commission
as well as some industrialists.
In this context, Autorité de Régulation des Telecommunications (ART) decided to conduct a study to identify the issues facing the migration of IPv4 to IPv6, in particular, the strategies of various players covering the entire value chain for services and Internet network equipment: equipment manufacturers, operators, ISPs, IP technology user companies, software publishers, etc. In particular, the study focussed on the issues of migration from a regulatory and competition viewpoint and its impact on the networks and telecommunications services markets using the IP protocol.
1 IPv6: well suited to the massive diffusion of Internet
1.1 IPv4: certain foreseeable limits - IPv6: strengths to support growth
IPv4: restricted addressing space with uneven geographic distribution
Today, much of this stock of IPv4 addresses has been used. And, while close to 47% of addresses have not been assigned (of the total stock of addresses), the geographic distribution is very uneven. The addresses allocated (intended for use by a regional register or pre-RIR organisations) represent most of the stock and are intended primarily for the American region at the expense of Asia, despite its major development potential (China, India). It is also important to note that of all available IPv4 addresses, 53% have been allocated directly to organisations (primarily American), before the appearance of Regional Internet Registries (RIR), which therefore do not control them.
Thus, if we take into account these pre-RIR organisations, we estimate that, by the end of 2001, 74% of addresses was allocated to North America, 17% to Europe and 9% to Asia.
In addition to the still strong organic growth of Internet around the world (and especially in Asia where the potential for growth is very strong and address resources very small), many new applications which consume IP addresses are expected to develop :
- mobile services based on GPRS, then UMTS
- high-speed access and "always on" mode
- on-line electronics and communicating vehicles
- home automation applications and sensor networks
Initially intended for non-commercial use, IPv4 was not designed to perform the QoS functions expected of it today, nor for auto configuration, Multicast functions or security, which are essential for modern commercial Internet. Solutions have been found to provide these functions, weighing down the protocol with extra layers, and to artificially extend the lifetime of the stock of addresses (NAT), rendering routing tables much more complex.
This management of the address shortage is reflected in the drastic IPv4 address attribution policies applied by the RIR. Moreover, the generalised use of NAT helps delay the shortage, but renders network management more complex and holds back the development of real-time and P2P applications.
Thus, all other things considered equal, we can expect the stock of IPv4 addresses to run out by 2010.
Despite the various technical advantages offered by IPv6 which we describe below, the most important, according to experts, is its large addressing space, which will help us meet the needs created by new "always on" applications and restore the use of the end-to-end mode which is the principal advantage offered by IPv6 at the application level. The other technical advantages, while real, are currently just interesting potentials, and not the main strength of IPv6:
- Hierarchical addressing to optimise routing
- Auto configuration
- Native IPSec
- Multicast
- Mobile IPv6
IPv6 can also better manage the Quality of Service (QoS), although this is not yet significant. As a general rule, we consider that initially, the QoS will be managed way under IPv6 as it is today under IPv4.
1.2 The IPv6 standard is stable
To date, the IPv6 core, essential for its operation, is considered to be stable by most specialists. Considering that IPv4 was used before all of its specifications were totally stable, we can deem IPv6 suitable for "commercial" use. What’s more, existing "construction sites" concern primarily the specific characteristics of IPv6, new aspects with respect to IPv4 :
- Management of global DNS under IPv6
- Mobile IPv6
- Auto configuration
- Flow Label field
Basically, the construction sites still in progress under IPv6 will have no negative impact on its development, since the protocol is complete enough to operate in a commercial mode as well as IPv4, and any new aspects will not affect this operation.
Telecoms equipment manufacturers provide the hardware required to route data on IP networks: primarily routers. Routers are currently designed to route data packets using IPv4. The use of IPv6—a different packet format—would initially require that these routing infrastructures be upgraded. These same telecoms equipment manufacturers also provide Internet access equipment: in particular CPE and mobile terminals.
On the consumer computer market, the operating system (OS) makes possible the use of IPv6. The OS produced by Microsoft, absolute leader on the market are almost ready for IPv6: Windows XP is already ready and will use IPv6 by default by end 2002 (for now, it is available but must be activated), and other Windows products will be ready before the end of 2002. The fact that the market leader is moving towards IPv6 is an essential element: if the most popular OS is IPv6 compatible, one of the major stumbling blocks for the transition will have been removed.
Backbone operators may create a bottleneck in the transition to IPv6: if they cannot guarantee data transit in IPv6, other operators and ISPs will be forced to use encapsulation techniques in IPv4 to send the data between remote nodes. These techniques may suffice in an initial deployment stage. However, large-scale tunnel management will not be viable; backbone operators will have to migrate their network core routers to IPv6 when justified by demand.
In Europe, leading equipment manufacturers specialising in mobility—and cellular networks in particular—are working actively in IPv6. Ericsson (which owns Telebit, a pioneer in IPv6 routers) and Nokia offer IPv6-compatible routers. These players consider that mobile Internet will be an impetus for growth for IP and that IPv6 will be needed to develop attractive services. They see 3G as a real opportunity for IPv6. What’s more, the IPv6 technological leap has been identified by the 3GPP in release 5 of the UMTS standard.
The WLAN market is seeing definite development in both the area of corporate networks and, more recently, in public access networks. WLANs can be used to connect both PDAs and PCs. Still, the expected number of terminals connected using these technologies is much lower than expected for mobile telephony. Thus, the development of WLANs will not have a major impact, at least initially, compared to mobile telephony. Still, more and more demand is felt today for the use of WLAN technologies on public networks. If this type of use were to be generalised, the arrival of Mobile IPv6 could be accelerated significantly; the attitude of regulators will be determining in coming months as to the use of WLAN technologies on public networks.
Telecoms equipment manufacturers have clearly identified IPv6 as an impetus for growth: IPv6 will accelerate equipment replacement (terminals and access networks). Plus, for equipment manufacturers, IPv6 will open up new markets :
- via on-line electronics: consumer equipment manufacturers
- through the intrusion of IP in sectors other than computing and telecommunications: IP equipment manufacturers
As part of the development of permanent terminal connections (in particular, development of broadband access, 3G), IPv6 will provide operators with a pool of addresses letting them offer their clients quality service, without complex management (NAT). Network management is simplified overall, mainly thanks to hierarchical addressing and auto configuration functions. In offering better service, we might imagine that operators will be able to increase prices for some parts of their offer. And, since network management is less costly, some equipment manufacturers and operators believe they will be able to increase margins.
The costs generated by the move to IPv6 for ISPs are
marginal, at least in terms of hardware: router software upgrades are
often free. On the other hand, if ISPs manage both IP versions on the
same network, management complexity may be felt, although the extra cost
of hardware remains marginal. The primary costs identified by access providers
are the human ones: personnel know how to use IPv4. IPv6 has new peculiarities
and the technicians will have to adapt to them. Plus, they will have to
be able to manage both IP versions simultaneously for a long period of
time.
The MPLS or Dual Stack architecture seems well suited for ISPs as it will
offer a fair degree of flexibility initially in response to client demand
for IPv6 access: ISPs can open IPv6 access case by case without having
to migrate the entire network.
Today, the main concern of ISPs, who operate on a particularly competitive
market currently in a concentration phase, is to stabilise their position
and find the best model allowing them to find equilibrium under IPv4.
Therefore, their main concern is not to branch out by offering IPv6, for
which demand has not yet been identified. The development of highspeed
access via ADSL for example, may trigger demand for IPv6 access for ISPs.
In Japan, ISPs (IIJ, NTT) are making their first commercial Ipv6 access
offers on ADSL.
As we have just mentioned, the abundance of IPv6 addresses
is an opportunity which will allow many ISPs to reposition themselves
on a more open market, offer new services, improve or simplify existing
services and more easily manage their networks.
Thus, the provision of Internet access services under IPv6 will fluidify
competition between ISPs once, for example, the auto configuration system
allows companies to change suppliers at a lower cost, since network renumbering
becomes automatic, but may also reduce the extent of their services.
We need to distinguish between two types of IP technology
companies: those using the protocol for their communications and networks
(Intranet, Extranet), and those which, while not players on the "traditional"
Internet market, may find opportunities to use IPv6 in new applications,
or as a substitute for existing applications.
These two aspects may be present in a single company, with different
attitudes with respect to IPv6 depending on the context.
The first—users of IP corporate networks—will not necessarily see
any need to move to IPv6. While they may recognise the potential advantages
of IPv6, their information system managers, as a whole, don’t see
any urgency. In the current context, their priority is to extend the
life of their investments already made under IPv4. This is the case
of most large companies who consider that, for now, IPv6 doesn’t really
offer anything that IPv4 doesn’t.
At the same time, certain companies not using IP, or not originally
oriented to the TIC, see an opportunity in IPv6; the Aeronautics sector,
for example, is actively examining the new IPv6 protocol. Players
see in it the possibility to move to IP. Plus, the abundance of addresses
could allow the development of new applications such as monitoring
of aircraft subassemblies, maintenance and on-board Internet. Carmakers,
while considering that their in-house networks can remain under IPv4,
see the possibility for new applications based on the Internet connected
vehicle using IPv6 in conjunction with cellular technologies. Appliance
manufacturers also foresee home automation applications.
Thus, major corporations are relatively neutral—or even indifferent
(on the part of IT departments)—with regard to IPv6, whereas R&D
departments see a number of opportunities in the field of communicating
objects. In any case, all are interested in being informed, although
sometimes that information isn’t always readily available.
3 Factors triggering the move to IPv6
3.1 Besides the shortage of IPv4 addresses, other front-line factors: mobile (GPRS and 3G) and nomad services (WLAN)
Besides the shortage of IPv4 addresses, the first
factor influencing the move to IPv6—expected in coming years (cf.
above)—other front-line factors were also highlighted in this study
:
The emergence of mobile data services, particularly in Japan with DoCoMo’s i-mode and other services offered by competitors, and in Europe with the success of SMS, leads us to wonder about the influence of this market on the introduction of IPv6. Not only that, but the arrival of new generations of network technologies should lead to a proliferation of on-line mobile terminals: first GPRS, whose first commercial corporate services are already offered by most GSM operators in western Europe, then 3G (UMTS and CDMA 2000) in which colossal investments are being made in Europe and whose first commercial service was opened in Japan by DoCoMo in October 2001.
Even though, initially, mobile GPRS and UMTS systems use IPv4, version IPv6 seems a major consideration for mobile network operators, in that it makes it possible to assign a permanent IP address to every on-line mobile terminal. Indeed, GPRS introduces the concept of "always on", that is, a permanent connection to the IP data network, even when the user is inactive. In the long run, sector players consider IPv6 to be an inevitable development for mobile networks.
Today, there is more and more demand for the use of WLAN technologies on public networks. If this type of use were to be generalised, it could considerably speed up the arrival of Mobile IPv6; the attitude of regulators will be a determining factor in coming months with regard to the use of WLAN technologies on public networks.
Native mobility management using IPv6 and its straightforward solutions simplify the mobility management of a terminal on a network (auto configuration, automatic renumbering) offering obvious advantages for this type of technology and making IPv6 a particularly attractive solution for mobility management in heterogeneous networks. This would be particularly true for mobile terminals via a WLAN, then on 3G networks: total and transparent mobility for the user.
3.2 The dynamics of Internet broadband access,
on-line electronics and sensor networks: second-string factors
After these front-line triggers, there are other reasons to migrate to IPv6:
The developing Internet broadband market could aggravate the address shortage under IPv4. Indeed, most high-speed access is in an always-on mode, meaning that the terminal is continuously on-line and therefore requires a fixed IP address. In practice, providers offering high-speed access (via ADSL for example) continue to offer dynamic addressing. However, the uses developing around these permanent connections mean that ISPs cannot apply the same modem/subscriber ratios as for a dial-up connection; these ratios can grow from 1/10 or 1/20 for dial-up to 1/2 or 1/4 for ADSL access. This accelerates the consumption of IP addresses. What’s more, if we examine the situation in Japan, we see that initial IPv6 deployments are not for mobile services as we might have expected, but for ADSL access (IIJ, NTT).
The development of on-line electronics is universally recognised as a potential lever for IPv6. Consumer electronics and appliances (such as TVs, cameras, etc.) will be more and more frequently connected to Internet: these might behave like terminals (television screens to surf on Internet, etc.) or servers (appliances, as home automation develops). Plus, the use of portable terminals like PDAs is expected to increase in the future and they will likely also be on-line. The need for IP addresses generated by these announced developments will likely make the move to IPv6 networks an absolute necessity, at least for the networks concerned by these applications.
Developments have already begun, particularly in Japan, where on-line appliances are expected to appear in 2003. Network gaming, growing the on-line gaming consoles market, is expected to explode in Asia in 2002.
The networking and connection of sensors via IP is an emerging technique for which experiments are currently underway (particularly in Japan). Many players see this accelerating the need for IP addresses: global addressing of "mega sensor networks", for weather forecasting, development of on-board sensor networks in cars and aircraft. The providers of military applications would also benefit from moving to IPv6: whether for "traditional" communications systems or for new equipment and troop monitoring systems. IPv6 offers real advantages. If these techniques do gain in popularity, they will increase the need for IP addresses and therefore push the move to IPv6.
4 IPv4 – IPv6: a long and inevitable coexistence
4.1 A limited risk of dividing up Internet
Even though IPv6 was designed to continue in the spirit of IPv4, with no real technological break, the new protocol is still different and interoperability between the two IPs is not natural. It would appear more judicious to speak of a transition and deployment rather than migration: this situation is not like Y2K with a sudden changeover, but rather a smooth and progressive transition. This means that the two standards will have to coexist for several years, and therefore interoperate. There are many transition scenarios and the IETF has foreseen the tools to support them, as well as their uses in the various transition phases.
Thus, the two standards could coexist for a relatively long time. We can consider that about ten years will be needed after the first IPv6 deployments, before IPv6 becomes used in the majority of applications. What’s more, there is no uniformity in this area: it is probable that ISPs will migrate before companies which first want to get their money’s worth from their applications developed under IPv4 before they invest in new techniques.
From a geographic viewpoint, the rate of entry in IPv6 of the various regions of the world will be quite different. Still, the geographic division (language barrier, habits of use) already exists, and the technical consequences of the coexistence of different standards in different zones should be minimal.
4.2 An opportunity to overhaul the distribution of
root DNS
The DNS server matches a domain name to an IP address. Currently, DNS servers operate under IPv4. While a few problems have been encountered when operating in IPv6, in Japan and in France, IPv6 DNS servers do work. The main problem will lie in ensuring IPv4/IPv6 interoperability for this function, in a transparent way during the domain name resolution process.
Currently, DNS servers are hosted by Internet governance
bodies. In France, Renater administers the first IPv6 DNS server. According
to experts, the main problem will not be technical and will not concern
"national" DNS servers: DNS organisation is hierarchical, and what is
missing today is the ability to manage the DNS at the top of this hierarchy:
root servers are not ready. The cause of this is "political" rather than
technical: a blockage by IANA and ICANN.
This means that international Internet authorities need to take a stand.
Indeed, it is in part by defining the hierarchy of the global IPv6 DNS
that we will build the Internet of tomorrow. Until a decision is taken,
we may see a division of the global DNS: one root DNS server in Asia,
another in Europe. Thus the arrival of IPv6 could be the opportunity to
review the domination of the United States in the current management of
the global DNS.
The arrival of IPv6 could lead to an intensification
of competition on existing markets linked to Internet access: by eliminating
the scarcity of IP address resources and lowering network management costs
for ISPs, barriers to entry could be removed; the auto configuration system
and terminal-to-terminal communication could help fluidify competition
between ISPs. New markets might appear, primarily for services (transition
solutions, access by new Internet access terminals, specific functions
offered by IPv6 for on-line gaming or videoconferencing services).
On the other hand, IPv6 might also create new bottlenecks if the
current restrictive address allocation mechanism isn’t made more flexible,
if a transit offer under IPv6 isn’t made available, if compatible systems
are late arriving or are limited. Dominant positions could be created
or strengthened on product (routers and operating systems) or service
markets, to the benefit of IPv6 backbone operators, or certain equipment
manufacturers.
IPV6 will be deployed primarily within a renewed regulatory
framework, by mid 2003 at the latest, marked by the harmonisation of the
system applicable to networks and electronic communications services.
On 21 March 2002, following the work done by the IPv6 Task Force, the
European Commission published a communication on next generation Internet,
proposing priority action for migration to the new IPv6 protocol.
The use of this protocol by operators won’t likely change the legal
qualification of networks and services, nor their system, but deployment
could be monitored based on information requested from operators.
The move to IPv6 will take place in the more global context
of Internet governance reform. Although a certain degree of opacity surrounds
the IPv6 address allocation rules because of a lack of visibility as regards
the adoption timetable and the number of participants in the process,
the future allocation rules will try to limit the risk of introducing
any inequality into competition.
We might note that with the definition of new IPv6 allocation policies,
efforts are being made to clarify the respective roles of the various
self-regulation bodies, reflecting a desire for greater transparency in
the adoption process. The draft agreement of 9 April 2002 between ICANN
and the RIR states that the RIR are responsible for allocating addresses
and developing policies for their region and that ICANN is responsible
for global coordination and resource management. The agreement recognises
the validity of the revision by the ASO of current addressing policies
and ICANN as the body responsible for assigning addresses to the RIR,
thereby ending IANA’s persistent incoherence in describing allocation
processes. However, regulators have no authority in the matter and governments
intervene only at the end of the chain in a consultative capacity.
The new IPv6 addressing policies, while reaffirming the
IP address management principles, will give greater importance to the
goal of aggregating rather than "preserving" addresses, and have introduced
a new principle of equity and impartiality for practices and policies.
The publication of the addresses assigned by the RIR, the absence of property
rights to assigned IP addresses and the supervision of transfers between
registers of the allocated addresses are all an effort to better manage
IPv6 addresses.
The current policy (defined temporarily in 1999) favours major players
and those who had positioned themselves early on experimental networks,
whereas the new policy works towards a rebalancing to the benefit of IPv4
players, although it does not totally eliminate any risk of discrimination
with respect to new "all IPv6" players. Discussions on the "IPv6
Address Allocation and Assignment Policy" project issued in December 2001
eased the access conditions to IPv6 addresses and take into account the
IPv4 client bases of requesters.
The recovery of unused addresses does not appear to be a truly effective means of delaying the shortage of IPv4 addresses: legal mechanisms for recovering addresses exist in theory although the RIPE do not appear to use them. A restrictive allocation and attribution policy for IPv4 addresses could delay the shortage of IPv4 addresses. This type of policy is stipulated by RFC 2050 and by the document RIPE 185 which limit the allocation and attribution of addresses to the specific needs of the address requester.
The new protocol should help ensure that ISPs eventually obtain a non-discriminating access offer to networks under IPV6. Interoperability of services seems assured by the existence of transition solutions and the availability of IPv4/IPv6-compatible routers and terminals. In case of need, not as yet seen, ETSI could theoretically intervene. The continuity of mobile multimedia service quality may lead to intervention on roaming agreements.
As for users, IPv6 has no impact on end-users’ choice of access operator or ISP, except if a practice were to be applied where ISP addresses were pre-installed in the terminals. As for free access to sites and content, ISPs hold the means to guarantee it, by offering content providers IPv6 access, i.e. in practice by purchasing dual-stack servers which can be used by all providers. Nevertheless, the auto configuration capability facilitates the pre-installation of addresses: depending on the agreements signed by terminal provider (and ISPs) with the third party, access to service platforms could be limited.
In its communication of 21 February 2002, the European Commission stated that IP addresses may be considered personal data under the EU legal framework. In principle, traffic data, such as IP addresses, must be erased or made anonymous upon completion of the transmission, barring a few exceptions. The new article L. 32-3-1 of the Post and Telecommunications Code establishes the principle of immediate erasure or anonymity of communication data, imposing this obligation on ISPs. The respect of the EU framework regarding the protection of privacy should also apply to RIPE and its databases.
A number of factors announce a forthcoming transition to IPv6 :
- the emergence of mobile data services based on GPRS and UMTS. However, this affirmation must be moderated. Initial versions of GPRS and UMTS are being deployed with IPv4
- Internet broadband access (fixed and nomad, such as WLAN) permanent address consumers
- on-line electronics which constitute potential leverage for the development of IPv6
Given the announced arrival of IPv6, we find players in various degrees of preparedness :
The positioning of these categories of players must be modulated by geographic zone: Asian players are overall the leaders in the move to IPv6.
The coexistence of IPv4 and IPv6 will be long and unavoidable. Therefore, it is important to speak of IPv6 deployment rather than migration.
Moreover, the arrival of IPv6 may bring into question American domination of global DNS management. In this respect, future decisions concerning the implementation of IPv6 root DNS will be crucial. With their DNS servers at the national level, Asia and Europe must immediately take a stand on these questions.
A number of regulatory concerns were raised by this study
:
Thus, if necessary, guidelines may provide the visibility desired by the market. These may concern the timetable of transit offers under IPV6 proposed by backbone operators or the principles to be respected by operators and ISPs to guarantee free access to the users of services and content. However, at present, the players questioned are more concerned with a favourable industrial environment than with specific legislative or regulatory intervention.
