IPv4 Shortage: Needs and Open Issues

Taking into consideration the IPv4 public address pool currently available at the Internet Assigned Numbers Authority (IANA), it is expected that the Regional Internet Registries (RIRs) will have no more public IPv4 addresses to allocate in the short term. At the time of writing, this anticipated date is mid-2012. See the IPv4 Address Report website www.potaroo.net/tools/ipv4/index.html for a thorough analysis of this issue, and an updated prediction. At the exhaustion date, ISPs will wind up with public address pools that cannot grow. They will have to make do with what they have currently got. They will enter an IPv4 address shortage management phase. It will not be possible to provide each customer with a unique public IPv4 address. On the other hand, offering only an access to the IPv6 Internet won't be satisfactory for the customers because a lot of services will remain IPv4-only accessible (a full IPv6 world requires universal adoption which is hard to achieve). This document analyses the main issues related to IPv4 Internet access in the context of public IPv4 address exhaustion.

So far, the current practice has been to give a unique IPv4 public address to each customer. Current designs assume the allocation of a global IPv4 address to the Customer Premises Equipment (CPE), whereas hosts connected to the CPE will be privately-addressed, meaning CPE devices activate a Network Address and Port Translator (NAPT) capability by default, sourcing IPv4 traffic based upon this sole global IPv4 address. In this context, the addresses that can be seen in any IP packets always refer to a unique customer. To cope with the IPv4 address exhaustion, this kind of practices is no more affordable. Therefore ISPs are bound to allocate the same IPv4 public address to several customers at the same time. All IPv4 address shortage mechanisms extend the address space in adding port information. They differ on the way they manage the port value. In this new context, an IPv4 address seen in an IP packet can refer to several customers. The port information must be considered as well, in order to be able to unambiguously identify the customer pointed by that shared address. In particular, the port information along with the address information, must eventually be taken into account by the routing infrastructure in order to correctly reach the intended destination. So far, two categories of solutions have been identified: (1) Solutions that propose the introduction of a NAPT function in the ISP network, denoted also as Carrier Grade NAT (CGN). The CGN is responsible for translating IP packets issued with private addresses to ones with publicly routable IPv4 addresses. (2) Solutions that avoid the presence of a CGN function, they allocate the same IP public address to several customers at the same time. They also allocate a restricted port range to each customer so that two customers with the same IP address have two different port ranges that do not overlap.

The purpose of sharing IPv4 addresses is to potentially increase the addressing space. A key parameter is the factor by which ISPs want or need to multiply their IPv4 public address space; and the consequence is the number of customers sharing the same public IPv4 address. The intention is not to replace IPv6. However, we should be very careful; whatever the network model deployed, applications and business will run on top of it. If we do not want to see IPv4 shortage mechanisms postpone IPv6 deployment, all Internet actors must adopt a voluntary position towards IPv6. It is expected that IPv6 traffic will take an increasing part during the next years to come, at the expense of IPv4 traffic. We should reach a safety point in the future, where the number of IPv4 public addresses, in use at the same time, begins decreasing. From an ISP point of view, the multiplicative factor must be enough to survive until this occurs for its own customers. Most likely, a one digit factor (less than 10) should be sufficient, and it should be pointless to go beyond. Whereas the potential is huge, -if we allow to each customer (one IP address, 1000 ports) we multiply by 64 the total IPv4 address space- trying to devise solutions that can increase the IPv4 space by a significantly bigger factor might be seen as an incentive to postpone again and again IPv6 deployment.

At the time of IPv4 address exhaustion in the RIRs, ISPs will have to manage public address pools that cannot grow (at least from the RIRs). Concretely, they will have to decide to whom they allocate shared addresses and to whom they allocate unique addresses, to the extent of the availability of addresses. Many policies can be envisaged, taking into account parameters such as: old vs. new customers, user profile, access type, geographic considerations, unique address as the privileged choice, shared address as the privileged choice, etc. An important issue is whether shared and unique addresses will be differently charged. Care must be taken when considering the ratio that reflects the number of customers who will share a given global IPv4 address, not only to preserve some flexibility on the global address space that is left, but also to make sure that the ISP can adequately serve customer's requirements, without degrading the services they have subscribed to. ISPs can adjust the volume of IPv4 public addresses available playing on the balance between shared and unique allocations. To increase the public IPv4 address pool: increase the number of customers with shared address; increase the ratio of customers per shared address. To decrease the public IPv4 address pool: decrease the number of customers with shared address; decrease the ratio of customers per shared address.

Any IPv4 address shortage solution should make use, as much as possible, of the IPv6 transport capabilities available, in order to increase the IPv6 packets traffic and to move forward from an IPv4-enabled ISP network towards an almost only IPv6-enabled ISP network. If it is not the case, the risk is to delay IPv6 deployments, in staying on a pure dual-stack attitude for ever, similar to the ships in the night routing approach, where the protocols independently live their own lives. The IPv4 in IPv6 tunnels, and/or the translation NAT464 should be favored. However, increasing the number of IPv6 packets does not automatically mean IPv6 is being generalized, if the main purpose of these packets is to carry IPv4 information. This is very similar to what occurred with ATM, especially in European countries, where ATM cells have heavily been used to convey IPv4 packets in the backhaul networks, but have never been used for end-to-end communications. If the percentage of end-to-end IPv6 traffic significantly increases, so that the volume of IPv4 traffic begins decreasing, then the number of IPv4 sessions will be decreasing. The smaller the number of current sessions per customer is, the higher the number of customers under the same IPv4 public address can be, and consequently, the lower the number of IPv4 public addresses is needed. Hence, the pressure on IPv4 address shortage would be relaxed, because one IPv4 public address would be able to potentially serve more customers. However, this effect will only occur for customers who have both an IPv6 access and a shared IPv4 access. This would advocate the strategy to systematically bound a shared IPv4 access to any IPv6 access. Furthermore, some public IPv4 addresses will be required to connect IPv4 and IPv6 realms, through IPv4 IPv6 translators, for the sake of global reachability.

All IPv4 address shortage solutions will be confronted to the hereafter listed issues. It is valuable to assess solutions with all these issues, to check to what degree they are concerned, how they handle each issue, and to what extent they resolve the pending problems.

The network addressing capability is the level of flexibility the network has to configure customers' devices, either with a unique address, or with a shared IPv4 address. It can be assessed through the following considerations: Is it possible to configure any customer's device with a shared address, regardless his location and his history? Is it possible to configure any customer's device with a unique public address, regardless his location and his history? Is it straightforward to switch, for any customer, from a shared address to a unique public address, and vice versa? What is considered here is not the policy decision to allocate a unique or a shared address, but indeed the network capability to enforce such address management schemes.

In any kind of solutions, the number of current sessions per customer has, de facto, to be limited in some way. Therefore, the number of current sessions per customer is a limit to take into account in any architectural dimensioning. The degree of fairness -balanced distribution of sessions between customers-, should be assessed. Means to prevent against traffic loss (due to the limitation in number of sessions) should be evaluated and proposed. The importance of this issue may be greatly reduced if the multiplicative factor is very small (e.g. 4). As for the current usage of ports, several hundreds per customer seems current practice, although several thousands may be not unusual with some P2P applications (e.g. BitTorrent). The impact of the dynamicity of the sessions should also be considered, especially as far as performance is concerned.

As a matter of fact, private IPv4 addresses (as defined in ) belong to a finite space which may rapidly raise overlapping issues as both the number of customers and the number of services that can be subscribed by these customers increase. As a consequence, some ISPs use Virtual Private Networks (VPNs) such as to allow reusing the same private addresses several times with no routing overlaps. This brings a lot of complexity in network configuration and management. It has been suggested to make the 240/4 block available for private addressing . This address block, formerly designated as "Class E", is still noted as being reserved in the IANA IPv4 address registry. If it were reassigned for private addressing that would yield around 268 millions extra private addresses. However, many current implementations of the TCP/IP protocol stack do not allow the use of the 240/4 block. This is a severe blocking point for a lot of existing devices: CPE, NAT or routers. This issue will only be solved when the vendors' implementations accept the (240/4) addresses. Another suggestion is to reserve some public blocks (typically three or four /8) only for internal usage. So far, there has been no consensus upon this proposal. What needs to be considered is to what extent the IPv4 sharing solutions make use of IPv4 private addresses.

Any claimed solution to solve the IPv4 address shortage should be able to deliver the IP connectivity services to a large amount of customers, this limit should be evaluated.

The impact on the Information System platforms and applications handling the administrative and technical information to control the activation of services, and to manage the customer profiles, should be evaluated and assessed. Common practice used to rely upon the global IPv4 address assigned to a CPE device for customer identification purposes. The forthcoming address depletion therefore encourages ISPs to revisit their customer identification schemes since global IPv4 addresses will be shared amongst several customers. This clearly advocates for an IPv6-based customer identification scheme and thus impacts the way customer-specific management policies are enforced. The possibility to give either a unique or a shared address, coupled or not with an IPv6 address, could yield several types of customers to deal with in the IS: IPv4 unique only, IPv4 shared only, IPv4 unique + IPv6, IPv4 shared + IPv6, IPv6 only. The way the solution tries to possibly alleviate or simplify customer profile handling, should be evaluated and assessed.

There is a potential danger for the following types of applications: Applications that establish inbound communications Applications that carry address information in their payload Applications that carry port information in their payload Applications that use fixed ports (e.g. well known) Applications that do not use any port (e.g. ICMP) Applications that assume the uniqueness of customers' addresses (e.g. IP address as identifier) Applications that explicitly prohibit twice the same address to access to a resource at the same time Current applications already implement some mechanisms in order to circumvent the presence of NATs (typically CPE NATs): ALGs Port Forwarding UPnP IGD NAT Traversal It should be considered to what extent these mechanisms can still be used with IPv4 shortage mechanisms put in place. Impact on existing services: Will this service work as usual? Will this service work but with a degradation? What level of degradation? Will this service not work at all? What modifications are needed if any? Impact on future services: What new constraints are to be taken into account to devise new services?

The ISP can offer walled garden services along with Internet services. The ISP may want these flows not to traverse the IPv4 shortage facilities put in place (for instance, all the IPv4 traffic doesn't have to be processed by a CGN facility, for various reasons that are mostly ISP policy-specific and which include -but are not necessarily limited to- performance considerations, service-specific forwarding policies). It should be clear how these IPv4 flows can bypass the IPv4 shortage facilities and how they can be handled by the corresponding service platform/gateway. However, the best practice seems to rapidly migrate these services from IPv4 to IPv6.

The introduction of port consideration to route packets to their final destinations may have an impact on the current routing infrastructure: on the architecture, the IGP and EGP configuration, the addressing configuration, and on routers performances. The introduction of new nodes that cannot be circumvent could also yield non optimized routes, especially for communications between customers attached to the same ISP realm. It could also strongly modify the current flow distribution scheme among the different links and nodes.

When a packet is fragmented, the port information (either UDP or TCP) will only be present in the first fragment. The other fragments will not bear the port information which is necessary to a correct treatment up to the destination. Appropriate solutions should be evaluated.

IPv4 shortage mechanisms may require specific features in the CPEs. Some CPEs are ISP branded. CPE devices are strategic not only because of their large number, but also because of the advanced capabilities they support, and which include (but are not limited to), firewalling, QoS and self-configuring abilities. The impact on existing CPE devices should be carefully evaluated, taking into account: features needed, required modifications, availability. This issue is not specific to ISP branded CPEs. CPE behavior should be particularly specified by any solution claiming to solve the IPv4 address exhaustion problem.

The possible degradation of end-to-end performances (e.g. delay) experienced in the context of IPv4 shortage solutions should be evaluated.

The impact on QoS mechanisms should be investigated. In particular the ability to classify traffic in order to apply differentiated treatments could be hindered by the fact that an IPv4 address is shared among several users, possibly in a dynamic way.

The introduction of new nodes/functions, specifically where the port information is managed, can create single points of failure. Any IPv4 shortage solution should consider the opportunity to add redundancy features in order to alleviate the impact on the robustness of the IP connectivity service. Additionally, load balancing and load sharing means should be evaluated. The ability of the solution to allow hot swapping from a machine to another, in minimizing the perturbations, should be considered.

Additional data related to port information should be stored and maintained by the ISP equipment. As an example, a set of entries (e.g. session states, binding entries) are to be instantiated and maintained. The amount of data to be stored, the way the entries are instantiated and managed, should be documented. The number of port related entries in the ISP equipment can have a significant impact on performance, scalability, and solution cost.

It should be assessed if a customer with a shared address can receive multicast packets and source multicast packets. Particularly, impact on IGMP should be identified and solutions proposed. Because of the presence of several end user devices with the same IP address, membership to multicast groups should be evaluated and enhancement should be proposed if required. Besides the membership issue, building multicast trees may be impacted. This impact should be assessed and alternatives proposed.

Owing to the deployment of a Mobile-IP architecture, a mobile terminal continues to access its connectivity service when visiting a Foreign Network. In order to avoid traffic loss, it is recommended to use the home address (HoA), and not the care-of address (CoA), to reach that mobile terminal. A dedicated entity called HA (Home Agent) is responsible for routing the traffic according to the binding table it maintains. This table includes in particular the association between the HoA and CoA. A Foreign Agent (FA) can optionally be deployed in the visited network. If an IP address is shared (in the home network or/and in the visited network), HA or FA must be updated so as to take into account the port information to achieve its operations (i.e. relay traffic destined to HoA to the current CoA). The way proposed solutions deal with Mobile-IP mechanisms should be identified and assessed.

In the current deployments, end-users are used to configuring their CPEs in order to control the traffic entering/exiting their home LAN. Examples of these facilities are: port forwarding or firewall rules. The availability of these facilities offer to the end-users should be considered in the context of IPv4 address exhaustion solutions. Degradation compared to the current practice should be assessed.

ISPs deploy a set of tools and applications for the management of their infrastructure, especially for supervision purposes. Impact on these tools should be evaluated and solutions proposed when required. Particularly, means to assign IP connectivity information, means to monitor the overall network, to assess the reachability of devices should be specified. In this context, impact on tools (e.g. ICMP-based) to check the reachability of network nodes should be evaluated.

ISPs can be required by governmental and/or regulation authorities to provide customer-specific information upon request.

Legal obligations require an ISP to provide the identity of a customer upon request of the authorities. Because one public IPv4 address may be shared between several users, the knowledge of the port number, along with the IP address, is mandatory to have a chance to find the appropriate user. The ISP must be able to provide the identity of a customer from the knowledge of the IPv4 public address and the port number.

This process is proactive, a given group of communications is replicated in real time towards a law enforcement agency. Typically, the point of replication is the first IP hop in the ISP network. Wiretapping techniques need to be transparent to the customer, so that the targeted customer cannot be aware of the interception.

A kind of blind attacks that can be performed against TCP relies on the attacker's ability to guess the five-tuple (Protocol, Source Address, Destination Address, Source Port, Destination Port) that identifies the transport protocol instance to be attacked. Document describes a number of methods for the random selection of the client port number, such that the possibility of an attacker guessing the exact value is reduced. With shared IPv4 addresses, the port selection space is reduced. Intuitively, assuming the port range is known, the smaller the port range is, the more predictable the port choice is. Any solution to solve IPv4 address shortage should specify how port randomization is impacted and what alternative means to bypass the issue are.

Some types of attacks that have an impact on a targeted IPv4 public address, could see their effects increased by the number of customers who share this address. For example, if a given address that has, deliberately or not misbehaved, is consequently forbidden to access some resources, the whole set of customers who share this address will be impacted.

Even if IPSec is not deployed for mass market, impacts of solutions based on shared IP addresses should be evaluated and assessed. proposes a solution to solve issues documented in . The applicability of in the context of shared IP address should be evaluated.

We are grateful to Christian Jacquenet, Iain Calder, and Marcelo Bagnulo, for their helpful comments and suggestions for improving this document.

There are no IANA considerations. Note to RFC Editor: this section may be removed on publication as an RFC.

Security considerations are discussed in the Security section