Operations Area Expiration date: November 28, 1992 Internet Draft A Proposal for A Global Internet Addressing Scheme Daniel Karrenberg Bernhard Stockman May 28, 1992 Status of this memo. This document is an Internet Draft. Internet Drafts are working documents of the Internet Engineering Task Force (IETF), its Areas, and its Working Groups. Note that other groups may also distribute working documents as Internet Drafts. Internet Drafts are draft documents valid for a maximum of six months. Internet Drafts may be updated, replaced, or obsoleted by other documents at any time. It is not appropriate to use Internet Drafts as reference material or to cite them other than as a ``working draft'' or ``work in progress.'' Please check the 1id-abstracts.txt listing contained in the internet-drafts Shadow Directories on nic.ddn.mil, nnsc.nsf.net, nic.nordu.net, ftp.nisc.sri.com, or munnari.oz.au to learn the current status of any Internet Draft. Abstract This is a proposal for a hierarchically based assignment of future IP addresses by distributed control of segments of the IP address space. 1. Introduction There are currently several proposals under review to cope with the exhaustion of IP address space and routing table growth until the Internet architecture can be extended to deal with the problems of rapid growth [CIDR]. All proposals use some form of aggregation of addresses to keep routing table size within the bounds of current technology. Also the Internet is becoming more and more global. This suggests that the procedure for assignment of addresses should be managed in a distributed way, rather than centrally as is the case now. Delegated registries will have to be set up in a hierarchical manner. The need for such a delegation of authority has been expressed in RFC 1174. These developments make a discussion about the distribution of the remaining IP address space necessary. Also consensus will have to be developed on all levels about the implementation of the delegated registries. This document is intended to serve as a basis for that discussion. 2. Delegated Registries Technically it can be argued that a two level hierarchy of Internet registries is sufficient: The top level registry and the service providers. Nonetheless the establishment of registries per geographic region and where desirable for smaller areas below that will be necessary. This will allow additional savings in routing table size by using larger aggregations. More importantly it will ensure a fair distribution of the address space and relyable registration of assignments. A regional registry will have more knowledge about the situation in the particular region than a global registry and thus be much better suited to evaluate the appropriateness of requests for address space. For the same reason regional registries are better suited to keep track of the registration information as it changes over time. Regional registries are also well suited to establish a system of checks and balances among themselves in order to ensure a globally fair distribution of the address space. Timely and open reporting about all assignments is essential for this as it is for proper operation of the Internet at large. A quick process of building global and regional consensus about a distributed Internet registry is needed. As soon as regional consensus is established and a regional registry set up, a suitable part of the address space should be assigned to the regional registry. Service providers which are not covered by a regional registry should receive only small allocations from the global registry in order to encourage the setup of regional registries. 3. Distribution of the Remaining IP Address Space A scheme for distribution of the IP address space must be fair on all levels from global to local while allowing for the level of aggregation needed to cope with routing table growth. While sequential allocation of all reasonable requests on a first come first served basis is the fairest scheme, it does not allow aggregation at all. The largest gains from aggregation can be realised by assigning the biggest possible blocks of addresses. In an environment where the amount of address space needed per region and per service provider is not known a priori this cannot be done in a fair manner. Any viable scheme must strike a balance between these extremes. The authors did not have access to historical data about IP address space assignment. Consequently the proposed scheme cannot be based on any prediction of growth rates in different regions. From recent allocations of the DDN NIC and a straightforward assumption of the allocation scheme we deduce that the class B and C address space is currently allocated roughly as follows: Class B 128.0-160.255 assigned 161.0-191.255 free Class C 192.0.0-192.255.255 assigned 193.0.0-223.255.255 free This does not take into account past block allocations which are only partially used. We propose to assign the remaining class B and C addresses to the regional registries in blocks containing the same number of host addresses. This will help to prevent globally suboptimal choices of network class caused by uneven supply. A somewhat arbitrary but straightforward choice is blocks of 255 Bs and 255x255 Cs. Assuming that this scheme can be implemented quickly it means that about 30 blocks of Bs and 30 blocks of Cs will be available before address space runs out. We believe this provides adequate granularity for a globally fair distribution of Internet addresses. On the other hand the impact on routing table size by sub-optimal aggregation is difficult to assess. An upper bound would be a large region which eventually uses 10 such blocks and has sufficiently underestimated the needs of a particular service provider resulting in this provider to get assignments from all 10 blocks. This very unlikely case will produce a one order of magnitude inefficiency on the lowest level of aggregation. On the global level the optimal number of aggregations per address class can be guessed at 5. The proposed scheme will produce 30. While this represents a factor of 6 the absolute numbers are low enough to still be extremely efficient. References: [RFC1174] IAB Recommended Policy on Distributing Internet Identifier Assignment. Vint Cerf. August 1990. [CIDR] Supernetting: an Address Assignment and Aggregation Strategy Vince Fuller, Tony Li, Jessica Yu Kannan Varadhan. April 23, 1992 Authors: Daniel Karrenberg RIPE NCC Kruislan 407 Amsterdam, Netherlands Phone +30 20 592 5065 Email: Daniel.Karrenberg@ripe.net Bernhard Stockman Royal Institute of Technology Stockholm, Sweden Phone: +46 8 790-6519 Email: boss@ebone.net Operations Area Expiration date: November 28, 1992 Internet Draft