RFC1338: Supernetting: an Address Assignment and Aggregation Strategy

Download in PDF format Download in text format

Obsoleted By:  RFC1519
Related keywords:  (internet address) (Routing)

Network Working Group                                         V. Fuller
Request for Comments: 1338                                      BARRNet
                                                                  T. Li
                                                                  J. Yu
                                                            K. Varadhan
                                                              June 1992

      Supernetting: an Address Assignment and Aggregation Strategy

Status of this Memo

   This memo provides information for the Internet community. It does
   not specify an Internet standard.  Distribution of this memo is


   This memo discusses strategies for address assignment of the existing
   IP address space with a view to conserve the address space and stem
   the explosive growth of routing tables in default-route-free routers
   run by transit routing domain providers.

Table of Contents

   Acknowledgements .................................................  2
   1.  Problem, goal, and motivation ................................  2
   2.  Scheme plan ..................................................  3
   2.1.  Aggregation and its limitations ............................  3
   2.2.  Distributed network number allocation ......................  5
   3.  Cost-benefit analysis ........................................  6
   3.1.  Present allocation figures .................................  7
   3.2.  Historic growth rates ......................................  8
   3.3.  Detailed analysis ..........................................  8
   3.3.1.  Benefits of new addressing plan ..........................  9
   3.3.2.  Growth rate projections ..................................  9
   4.  Changes to Inter-Domain routing protocols .................... 11
   4.1.  General semantic changes ................................... 11
   4.2.  Rules for route advertisement .............................. 11
   4.3.  How the rules work ......................................... 13
   4.4.  Responsibility for and configuration of aggregation ........ 14
   5.  Example of new allocation and routing ........................ 15
   5.1.  Address allocation ......................................... 15
   5.2.  Routing advertisements ..................................... 17
   6.  Transitioning to a long term solution ........................ 18

Fuller, Li, Yu, & Varadhan                                      [Page 1]
RFC 1338                      Supernetting                     June 1992

   7.  Conclusions .................................................. 18
   8.  Recommendations .............................................. 18
   9.  Bibliography ................................................. 19
   10. Security Considerations ...................................... 19
   11. Authors' Addresses ........................................... 19


   The authors wish to express their appreciation to the members of the
   ROAD group with whom many of the ideas contained in this document
   were inspired and developed.

1.    Problem, Goal, and Motivation

   As the Internet has evolved and grown over in recent years, it has
   become painfully evident that it is soon to face several serious
   scaling problems. These include:

        1.   Exhaustion of the class-B network address space. One
             fundamental cause of this problem is the lack of a network
             class of a size which is appropriate for mid-sized
             organization; class-C, with a maximum of 254 host
             addresses, is too small while class-B, which allows up to
             65534 addresses, is to large to be widely allocated.

        2.   Growth of routing tables in Internet routers beyond the
             ability of current software (and people) to effectively

        3.   Eventual exhaustion of the 32-bit IP address space.

   It has become clear that the first two of these problems are likely
   to become critical within the next one to three years.  This memo
   attempts to deal with these problems by proposing a mechanism to slow
   the growth of the routing table and the need for allocating new IP
   network numbers. It does not attempt to solve the third problem,
   which is of a more long-term nature, but instead endeavors to ease
   enough of the short to mid-term difficulties to allow the Internet to
   continue to function efficiently while progress is made on a longer-
   term solution.

   The proposed solution is to hierarchically allocate future IP address
   assignment, by delegating control of segments of the IP address space
   to the various network service providers.

   It is proposed that this scheme of allocating IP addresses be
   undertaken as soon as possible.  It is also believed that the scheme
   will suffice as a short term strategy, to fill the gap between now

Fuller, Li, Yu, & Varadhan                                      [Page 2]
RFC 1338                      Supernetting                     June 1992

   and the time when a viable long term plan can be put into place and
   deployed effectively.  It is believed that this scheme would be
   viable for at least three (3) years, in which time frame, a suitable
   long term solution would be expected to be deployed.

   Note that this plan neither requires nor assumes that already
   assigned addresses will be reassigned, though if doing so were
   possible, it would further reduce routing table sizes. It is assumed
   that routing technology will be capable of dealing with the current
   routing table size and with some reasonably-small rate of growth.
   The emphasis of this plan is on significantly slowing the rate of
   this growth.

   This scheme will not affect the deployment of any specific long term
   plan, and therefore, this document will not discuss any long term
   plans for routing and address architectures.

2.    Scheme Plan

   There are two basic components of this addressing and routing scheme:
   one, to distribute the allocation of Internet address space and two,
   to provide a mechanism for the aggregation of routing information.

   2.1.  Aggregation and its limitations

   One major goal of this addressing plan is to allocate Internet
   address space in such a manner as to allow aggregation of routing
   information along topological lines. For simple, single-homed
   clients, the allocation of their address space out of a service
   provider's space will accomplish this automatically - rather than
   advertise a separate route for each such client, the service provider
   may advertise a single, aggregate, route which describes all of the
   destinations contained within it. Unfortunately, not all sites are
   singly-connected to the network, so some loss of ability to aggregate
   is realized for the non simple cases.

   There are two situations that cause a loss of aggregation efficiency.

     o    Organizations which are multi-homed. Because multi-homed
          organizations must be advertised into the system by each of
          their service providers, it is often not feasible to aggregate
          their routing information into the address space any one of
          those providers. Note that they still may receive their
          address allocation out of a service provider's address space
          (which has other advantages), but their routing information
          must still be explicitly advertised by most of their service
          providers (the exception being that if the site's allocation
          comes out of its least-preferable service provider, then that

Fuller, Li, Yu, & Varadhan                                      [Page 3]
RFC 1338                      Supernetting                     June 1992

          service provider need not advertise the explicit route -
          longest-match will insure that its aggregated route is used to
          get to the site on a non-primary basis).  For this reason, the
          routing cost for these organizations will typically be about
          the same as it is today.

     o    Organizations which move from one service provider to another.
          This has the effect of "punching a hole" in the aggregation of
          the original service provider's advertisement. This plan will
          handle the situation by requiring the newer service provider
          to advertise a specific advertisement for the new client,
          which is preferred by virtue of being the longest match.  To
          maintain efficiency of aggregation, it is recommended that
          organizations which do change service providers plan to
          eventually migrate their address assignments from the old
          provider's space to that of the new provider. To this end, it
          is recommended that mechanisms to facilitate such migration,
          including improved protocols and procedures for dynamic host
          address assignment, be developed.

     Note that some aggregation efficiency gain can still be had for
     multi-homed sites (and, in general, for any site composed of
     multiple, logical IP network numbers) - by allocating a contiguous
     block of network numbers to the client (as opposed to multiple,
     independently represented network numbers) the client's routing
     information may be aggregated into a single (net, mask) pair. Also,
     since the routing cost associated with assigning a multi-homed site
     out of a service provider's address space is no greater than the
     current method of a random allocation by a central authority, it
     makes sense to allocate all address space out of blocks assigned to
     service providers.

     It is also worthwhile to mention that since aggregation may occur
     at multiple levels in the system, it may still be possible to
     aggregate these anomalous routes at higher levels of whatever
     hierarchy may be present. For example, if a site is multi-homed to
     two NSFNet regional networks both of whom obtain their address
     space from the NSFNet, then aggregation by the NSFNet of routes
     from the regionals will include all routes to the multi-homed site.

     Finally, it should also be noted that deployment of the new
     addressing plan described in this document may (and should) begin
     almost immediately but effective use of the plan to aggregate
     routing information will require changes to some Inter-Domain
     routing protocols. Likewise, deploying the supernet-capable Inter-
     Domain protocols without deployment of the new address plan will
     not allow useful aggregation to occur (in other words, the

Fuller, Li, Yu, & Varadhan                                      [Page 4]
RFC 1338                      Supernetting                     June 1992

     addressing plan and routing protocol changes are both required for
     supernetting, and its resulting reduction in table growth, to be
     effective.) Note, however, that during the period of time between
     deployment of the addressing plan and deployment of the new
     protocols, the size of routing tables may temporarily grow very
     rapidly. This must be considered when planning the deployment of
     the two plans.

     Note: in the discussion and examples which follow, the network+mask
     notation is used to represent routing destinations. This is used
     for illustration only and does not require that routing protocols
     use this representation in their updates.

     2.2.  Distributed allocation of address space

     The basic idea of the plan is to allocate one or more blocks of
     Class-C network numbers to each network service provider.
     Organizations using the network service provider for Internet
     connectivity are allocated bitmask-oriented subsets of the
     provider's address space as required.

     Note that in contrast to a previously described scheme of
     subnetting a class-A network number, this plan should not require
     difficult host changes to work around domain system limitations -
     since each sub-allocated piece of the address space looks like a
     class-C network number, delegation of authority for the IN-
     ADDR.ARPA domain works much the same as it does today - there will
     just be a lot of class-C network numbers whose IN-ADDR.ARPA
     delegations all point to the same servers (the same will be true of
     the root delegating a large block of class-Cs to the network
     provider, unless the delegation just happens to fall on a byte
     boundary). It is also the case that this method of aggregating
     class-C's is somewhat easier to deploy, since it does not require
     the ability to split a class-A across a routing domain boundary
     (i.e., non-contiguous subnets).

     It is also worthy to mention that once Inter-Domain protocols which
     support classless network destinations are widely deployed, the
     rules described by the "supernetting" plan generalize to permit
     arbitrary super/subnetting of the remaining class-A and class-B
     address space (the assumption being that classless Inter-Domain
     protocols will either allow for non-contiguous subnets to exist in
     the system or that all components of a sub-allocated class-A/B will
     be contained within a single routing domain). This will allow this
     plan to continue to be used in the event that the class-C space is
     exhausted before implementation of a long-term solution is deployed
     (there may, however, be further implementation considerations
     before doing this).

Fuller, Li, Yu, & Varadhan                                      [Page 5]
RFC 1338                      Supernetting                     June 1992

     Hierarchical sub-allocation of addresses in this manner implies
     that clients with addresses allocated out of a given service
     provider are, for routing purposes, part of that service provider
     and will be routed via its infrastructure. This implies that
     routing information about multi-homed organizations, i.e.,
     organizations connected to more than one network service provider,
     will still need to be known by higher levels in the hierarchy.

     The advantages of hierarchical assignment in this fashion are

     a)   It is expected to be easier for a relatively small number of
          service providers to obtain addresses from the central
          authority, rather than a much larger, and monotonically
          increasing, number of individual clients.  This is not to be
          considered as a loss of part of the service providers' address

     b)   Given the current growth of the Internet, a scalable and
          delegatable method of future allocation of network numbers has
          to be achieved.

   For these reasons, and in the interest of providing a consistent
   procedure for obtaining Internet addresses, it is recommended that
   most, if not all, network numbers be distributed through service

3.  Cost-benefit analysis

   This new method of assigning address through service providers can be
   put into effect immediately and will, from the start, have the
   benefit of distributing the currently centralized process of
   assigning new addresses. Unfortunately, before the benefit of
   reducing the size of globally-known routing destinations can be
   achieved, it will be necessary to deploy an Inter-Domain routing
   protocol capable of handling arbitrary network+mask pairs. Only then
   will it be possible to aggregate individual class-C networks into
   larger blocks represented by single routing table entries.

   This means that upon introduction, the new addressing plan will not
   in and of itself help solve the routing table size problem. Once the
   new Inter-Domain routing protocol is deployed, however, an immediate
   drop in the number of destinations which clients of the new protocol
   must carry will occur.  A detailed analysis of the magnitude of this
   expected drop and the permanent reduction in rate of growth is given
   in the next section.

   In should also be noted that the present method of flat address
   allocations imposes a large bureaucratic cost on the central address

Fuller, Li, Yu, & Varadhan                                      [Page 6]
RFC 1338                      Supernetting                     June 1992

   allocation authority. For scaling reasons unrelated to address space
   exhaustion or routing table overflow, this should be changed. Using
   the mechanism proposed in this paper will have the happy side effect
   of distributing the address allocation procedure, greatly reducing
   the load on the central authority.

   3.1.  Present Allocation Figures

      A back-of-the-envelope analysis of "network-contacts.txt"
      (available from the DDN NIC) indicates that as of 2/25/92, 46 of
      126 class-A network numbers have been allocated (leaving 81) and
      5467 of 16256 class-B numbers have been allocated, leaving 10789.
      Assuming that recent trends continue, the number of allocated
      class-B's will continue to double approximately once a year. At
      this rate of grown, all class-B's will be exhausted within about
      15 months.

Fuller, Li, Yu, & Varadhan                                      [Page 7]
RFC 1338                      Supernetting                     June 1992

   3.2.  Historic growth rates

      MM/YY     ROUTES                        MM/YY     ROUTES
                ADVERTISED                              ADVERTISED
      ------------------------                -----------------------
      Feb-92    4775                          Apr-90    1525
      Jan-92    4526                          Mar-90    1038
      Dec-91    4305                          Feb-90    997
      Nov-91    3751                          Jan-90    927
      Oct-91    3556                          Dec-89    897
      Sep-91    3389                          Nov-89    837
      Aug-91    3258                          Oct-89    809
      Jul-91    3086                          Sep-89    745
      Jun-91    2982                          Aug-89    650
      May-91    2763                          Jul-89    603
      Apr-91    2622                          Jun-89    564
      Mar-91    2501                          May-89    516
      Feb-91    2417                          Apr-89    467
      Jan-91    2338                          Mar-89    410
      Dec-90    2190                          Feb-89    384
      Nov-90    2125                          Jan-89    346
      Oct-90    2063                          Dec-88    334
      Sep-90    1988                          Nov-88    313
      Aug-90    1894                          Oct-88    291
      Jul-90    1727                          Sep-88    244
      Jun-90    1639                          Aug-88    217
      May-90    1580                          Jul-88    173

            Table I : Growth in routing table size, total numbers
                      Source for the routing table size data is MERIT

   3.3.   Detailed Analysis

      There is no technical cost and minimal administrative cost
      associated with deployment of the new address assignment plan. The
      administrative cost is basically that of convincing the NIC, the
      IANA, and the network service providers to agree to this plan,
      which is not expected to be too difficult. In addition,
      administrative cost for the central numbering authorities (the NIC
      and the IANA) will be greatly decreased by the deployment of this
      plan. To take advantage of aggregation of routing information,
      however, it is necessary that the capability to represent routes
      as arbitrary network+mask fields (as opposed to the current
      class-A/B/C distinction) be added to the common Internet inter-
      domain routing protocol(s).

Fuller, Li, Yu, & Varadhan                                      [Page 8]
RFC 1338                      Supernetting                     June 1992

   3.3.1. Benefits of the new addressing plan

      There are two benefits to be had by deploying this plan:

      o    The current problem with depletion of the available class-B
           address space can be ameliorated by assigning more-
           appropriately sized blocks of class-C's to mid-sized
           organizations (in the 200-4000 host range).

      o    When the improved inter-domain routing protocol is deployed,
           an immediate decrease in the number routing table entries
           followed by a significant reduction in the rate growth of
           routing table size should occur (for default-free routers).

   3.3.2. Growth rate projections

      Currently, a default-free routing table (for example, the routing
      tables maintained by the routers in the NSFNET backbone) contains
      approximately 4700 entries. This number reflects the current size
      of the NSFNET routing database. Historic data shows that this
      number, on average, has doubled every 10 months between 1988 and
      1991. Assuming that this growth rate is going to persist in the
      foreseeable future (and there is no reason to assume otherwise),
      we expect the number of entries in a default-free routing table to
      grow to approximately 30000 in two(2) years time.  In the
      following analysis, we assume that the growth of the Internet has
      been, and will continue to be, exponential.

      It should be stressed that these projections do not consider that
      the current shortage of class-B network numbers may increase the
      number of instances where many class-C's are used rather than a
      class-B. Using an assumption that new organizations which formerly
      obtained class-B's will now obtain somewhere between 4 and 16
      class-C's, the rate of routing table growth can conservatively be
      expected to at least double and probably quadruple. This means the
      number of entries in a default-free routing table may well exceed
      10,000 entries within six months and 20,000 entries in less than a

      Under the proposed plan, growth of the routing table in a
      default-free router is greatly reduced since most new address
      assignment will come from one of the large blocks allocated to the
      service providers.  For the sake of this analysis, we assume
      prompt implementation of this proposal and deployment of the
      revised routing protocols. We make the initial assumption that any
      initial block given to a provider is sufficient to satisfy its
      needs for two years.

Fuller, Li, Yu, & Varadhan                                      [Page 9]
RFC 1338                      Supernetting                     June 1992

      Since under this plan, multi-homed networks must continue to be
      explicitly advertised throughout the system (according to Rule#1
      described in section 4.2), the number multi-homed routes is
      expected to be the dominant factor in future growth of routing
      table size, once the supernetting plan is applied.

      Presently, it is estimated that there are fewer than 100 multi-
      homed organizations connected to the Internet. Each such
      organization's network is comprised of one or more network
      numbers.  In many cases (and in all future cases under this plan),
      the network numbers used by an organization are consecutive,
      meaning that aggregation of those networks during route
      advertisement may be possible. This means that the number of
      routes advertised within the Internet for multi-homed networks may
      be approximated as the total number of multi-homed organizations.
      Assuming that the number of multi-homed organization will double
      every year (which may be a over-estimation, given that every
      connection costs money), the number of routes for multi-homed
      networks would be expected to grow to approximately 800 in three

      If we further assume that there are approximately 100 service
      providers, then each service provider will also need to advertise
      its block of addresses.  However, due to aggregation, these
      advertisements will be reduced to only 100 additional routes.  We
      assume that after the initial two years, new service providers
      combined with additional requests from existing providers will
      require an additional 50 routes per year.  Thus, the total is 4700
      + 800 + 150 = 5650.  This represents an annual grown rate of
      approximately 6%.  This is in clear contrast to the current annual
      growth of 150%.  This analysis also assumes an immediate
      deployment of this plan with full compliance. Note that this
      analysis assumes only a single level of route aggregation in the
      current Internet - intelligent address allocation should
      significantly improve this.

      Clearly, this is not a very conservative assumption in the
      Internet environment nor can 100% adoption of this proposal be
      expected. Still, with only a 90% participation in this proposal by
      service providers, at the end of the target three years, global
      routing table size will be "only" 4700 + 800 + 145 + 7500 = 13145
      routes -- without any action, the routing table will grow to
      approximately 75000 routes during that time period.

Fuller, Li, Yu, & Varadhan                                     [Page 10]
RFC 1338                      Supernetting                     June 1992

4.    Changes to Inter-Domain routing protocols

   In order to support supernetting efficiently, it is clear that some
   changes will need to be made to both routing protocols themselves and
   to the way in which routing information is interpreted. In the case
   of "new" inter-domain protocols, the actual protocol syntax changes
   should be relatively minor. This mechanism will not work with older
   inter-domain protocols such as EGP2; the only ways to interoperate
   with old systems using such protocols are either to use existing
   mechanisms for providing "default" routes or b) require that new
   routers talking to old routers "explode" supernet information into
   individual network numbers.  Since the first of these is trivial
   while the latter is cumbersome (at best -- consider the memory
   requirements it imposes on the receiver of the exploded information),
   it is recommended that the first approach be used -- that older
   systems to continue to the mechanisms they currently employ for
   default handling.

   Note that a basic assumption of this plan is that those organizations
   which need to import "supernet" information into their routing
   systems must run IGPs (such as OSPF[RFC1267]) which support classless
   routes. Systems running older IGPs may still advertise and receive
   "supernet" information, but they will not be able to propagate such
   information through their routing domains.

   4.1.  Protocol-independent semantic changes

   There are two fundamental changes which must be applied to Inter-
   Domain routing protocols in order for this plan to work. First, the
   concept of network "class" needs to be deprecated - this plan assumes
   that routing destinations are represented by network+mask pairs and
   that routing is done on a longest-match basis (i.e., for a given
   destination which matches multiple network+mask pairs, the match with
   the longest mask is used). Second, current Inter-Domain protocols
   generally do not support the concept of route aggregation, so the new
   semantics need to be implemented mechanisms that routers use to
   interpret routing information returned by the Inter-Domain protocols.
   In particular, when doing aggregation, dealing with multi-homed sites
   or destinations which change service providers is difficult.
   Fortunately, it is possible to define several fairly simple rules for
   dealing with such cases.

   4.2.  Rules for route advertisement

     1.   Routing to all destinations must be done on a longest-match
          basis only.  This implies that destinations which are multi-
          homed relative to a routing domain must always be explicitly
          announced into that routing domain - they cannot be summarized

Fuller, Li, Yu, & Varadhan                                     [Page 11]
RFC 1338                      Supernetting                     June 1992

          (this makes intuitive sense - if a network is multi-homed, all
          of its paths into a routing domain which is "higher" in the
          hierarchy of networks must be known to the "higher" network).

     2.   A routing domain which performs summarization of multiple
          routes must discard packets which match the summarization but
          do not match any of the explicit routes which makes up the
          summarization. This is necessary to prevent routing loops in
          the presence of less-specific information (such as a default
          route).  Implementation note - one simple way to implement
          this rule would be for the border router to maintain a "sink"
          route for each of its aggregations. By the rule of longest
          match, this would cause all traffic destined to components of
          the aggregation which are not explicitly known to be

   Note that during failures, partial routing of traffic to a site which
   takes its address space from one service provider but which is
   actually reachable only through another (i.e., the case of a site
   which has change service providers) may occur because such traffic
   will be routed along the path advertised by the aggregated route.
   Rule #2 will prevent any real problem from occurring by forcing such
   traffic to be discarded by the advertiser of the aggregated route,
   but the output of "traceroute" and other similar tools will suggest
   that a problem exists within the service provider advertising the
   aggregate, which may be confusing to network operators (see the
   example in section 5.2 for details). Solutions to this problem appear
   to be challenging and not likely to be implementable by current
   Inter-Domain protocols within the time-frame suggested by this
   document. This decision may need to be revisited as Inter-Domain
   protocols evolve.

   An implementation following these rules should also make the
   implementation be generalized, so that arbitrary network number and
   mask are accepted for all routing destinations.  The only outstanding
   constraint is that the mask must be left contiguous.  Note that the
   degenerate route mask is used as a default route and
   MUST be accepted by all implementations.  Further, to protect against
   accidental advertisements of this route via the inter-domain
   protocol, this route should never be advertised unless there is
   specific configuration information indicating to do so.

Fuller, Li, Yu, & Varadhan                                     [Page 12]
RFC 1338                      Supernetting                     June 1992

   Systems which process route announcements must also be able to verify
   that information which they receive is correct. Thus, implementations
   of this plan which filter route advertisements must also allow masks
   in the filter elements.  To simplify administration, it would be
   useful if filter elements automatically allowed more specific network
   numbers and masks to pass in filter elements given for a more general
   mask.  Thus, filter elements which looked like:


   would look something like:


   This is merely making explicit the network mask which was implied by
   the class-A/B/C classification of network numbers.

   4.3.  How the rules work

   Rule #1 guarantees that the routing algorithm used is consistent
   across implementations and consistent with other routing protocols,
   such as OSPF. Multi-homed networks are always explicitly advertised
   by every service provider through which they are routed even if they
   are a specific subset of one service provider's aggregate (if they
   are not, they clearly must be explicitly advertised). It may seem as
   if the "primary" service provider could advertise the multi-homed
   site implicitly as part of its aggregate, but the assumption that
   longest-match routing is always done causes this not to work.

   Rule #2 guarantees that no routing loops form due to aggregation.
   Consider a mid-level network which has been allocated the 2048
   class-C networks starting with (see the example in section
   5 for more on this).  The mid-level advertises to a "backbone" Assume that the "backbone", in turn, has been
   allocated the block of networks The backbone
   will then advertise this aggregate route to the mid-level. Now, if
   the mid-level loses internal connectivity to the network (which is part of its aggregate), traffic
   from the "backbone" to the mid-level to destination will
   follow the mid-level's advertised route. When that traffic gets to
   the mid-level, however, the mid-level *must not* follow the route

Fuller, Li, Yu, & Varadhan                                     [Page 13]
RFC 1338                      Supernetting                     June 1992 it learned from the backbone, since that would
   result in a routing loop. Rule #2 says that the mid-level may not
   follow a less-specific route for a destination which matches one of
   its own aggregated routes. Note that handling of the "default" route
   ( is a special case of this rule - a network must not
   follow the default to destinations which are part of one of it's
   aggregated advertisements.

   4.4.  Responsibility for and configuration of aggregation

   The AS which owns a range of addresses has the sole authority for
   aggregation of its address space.  In the usual case, the AS will
   install manual configuration commands in its border routers to
   aggregate some portion of its address space.  As AS can also delegate
   aggregation authority to another AS.  In this case, aggregation is
   done in the other AS by one of its border routers.

   When an inter-domain border router performs route aggregation, it
   needs to know the range of the block of IP addresses to be
   aggregated.  The basic principle is that it should aggregate as much
   as possible but not to aggregate those routes which cannot be treated
   as part of a single unit due to multi-homing, policy, or other

   One mechanism is to do aggregation solely based on dynamically
   learned routing information. This has the danger of not specifying a
   precise enough range since when a route is not present, it is not
   always possible to distinguish whether it is temporarily unreachable
   or that it does not belong in the aggregate. Purely dynamic routing
   also does not allow the flexibility of defining what to aggregate
   within a range. The other mechanism is to do all aggregation based on
   ranges of blocks of IP addresses preconfigured in the router.  It is
   recommended that preconfiguration be used, since it more flexible and
   allows precise specification of the range of destinations to

   Preconfiguration does require some manually-maintained configuration
   information, but not excessively more so than what router
   administrators already maintain today. As an addition to the amount
   of information that must be typed in and maintained by a human,
   preconfiguration is just a line or two defining the range of the
   block of IP addresses to aggregate. In terms of gathering the
   information, if the advertising router is doing the aggregation, its
   administrator knows the information because the aggregation ranges
   are assigned to its domain.  If the receiving domain has been granted
   the authority to and task of performing aggregation, the information
   would be known as part of the agreement to delegate aggregation.
   Given that it is common practice that a network administrator learns

Fuller, Li, Yu, & Varadhan                                     [Page 14]
RFC 1338                      Supernetting                     June 1992

   from its neighbor which routes it should be willing to accept,
   preconfiguration of aggregation information does not introduce
   additional administrative overhead.

5.    Example of new allocation and routing

   5.1.  Address allocation

   Consider the block of 2048 class-C network numbers beginning with (0xC0180000 and ending with (0xC01FFF00)
   allocated to a single network provider, "RA". A "supernetted" route
   to this block of network numbers would be described as
   with mask of (0xFFF80000).

   Assume this service provider connects six clients in the following
   order (significant because it demonstrates how temporary "holes" may
   form in the service provider's address space):

       "C1" requiring fewer than 2048 addresses (8 class-C networks)

       "C2" requiring fewer than 4096 addresses (16 class-C networks)

       "C3" requiring fewer than 1024 addresses (4 class-C networks)

       "C4" requiring fewer than 1024 addresses (4 class-C networks)

       "C5" requiring fewer than 512 addresses (2 class-C networks)

       "C6" requiring fewer than 512 addresses (2 class-C networks)

   In all cases, the number of IP addresses "required" by each client is
   assumed to allow for significant growth. The service provider
   allocates its address space as follows:

       C1: allocate 192.24.0 through 192.24.7. This block of networks is
           described by the "supernet" route and mask

       C2: allocate 192.24.16 through 192.24.31. This block is described
           by the route, mask

       C3: allocate 192.24.8 through 192.24.11. This block is described
           by the route, mask

       C4: allocate 192.24.12 through 192.24.15. This block is described
           by the route, mask

       C5: allocate 192.24.32 and 192.24.33. This block is described by

Fuller, Li, Yu, & Varadhan                                     [Page 15]
RFC 1338                      Supernetting                     June 1992

           the route, mask

       C6: allocate 192.24.34 and 192.24.35. This block is described by
           the route, mask

   Note that if the network provider uses an IGP which can support
   classless networks, he can (but doesn't have to) perform
   "supernetting" at the point where he connects to his clients and
   therefore only maintain six distinct routes for the 36 class-C
   network numbers. If not, explicit routes to all 36 class-C networks
   will have to be carried by the IGP.

   To make this example more realistic, assume that C4 and C5 are multi-
   homed through some other service provider, "RB". Further assume the
   existence of a client "C7" which was originally connected to "RB" but
   has moved to "RA". For this reason, it has a block of network numbers
   which are allocated out "RB"'s block of (the next) 2048 class-C
   network numbers:

       C7: allocate 192.32.0 through 192.32.15. This block is described
           by the route 192.32.0, mask

   For the multi-homed clients, we will assume that C4 is advertised as
   primary via "RA" and secondary via "RB"; C5 is primary via "RB" and
   secondary via "RA". To connect this mess together, we will assume
   that "RA" and "RB" are connected via some common "backbone" provider

   Graphically, this simple topology looks something like this:

Fuller, Li, Yu, & Varadhan                                     [Page 16]
RFC 1338                      Supernetting                     June 1992

                       C1 -- \         _ -  \       /
                           \     /             C7
                       C2  +----+                                 +----+ - \|    |                                 |    |  |    |  _ - _  |    |
                           |    | /  \ |    |
                       C3 -|    |/              C4               \|    | -   | RA |                                 | RB |   |    |___ - ___|    |
                          /|    |    |    |
                       C6  |    |               C5                |    | -  |    |                                 |    |  |    |                                 |    |
                           +----+                                 +----+
                              \\                                     \\ (C4) || (C4) || (C5) || (C5) ||  (C7) || (RB)    || (RA)    ||                                     ||
                               VV                                     VV
                     +--------------- BACKBONE PEER  BB ---------------+

   5.2.  Routing advertisements

   To follow rule #1, RA will need to advertise the block of addresses
   that it was given and C7.  Since C4 and C5 are multi-homed, they must
   also be advertised.

   Advertisements from "RA" to "BB" will be: primary    (advertises C4) secondary  (advertises C5) primary     (advertises C7) primary       (advertises remainder of RA)

   For RB, the advertisements must also include C4 and C5 as well as
   it's block of addresses.  Further, RB may advertise that C7 is

   Advertisements from "RB" to "BB" will be: secondary  (advertises C4) primary    (advertises C5) primary       (advertises remainder of RB)

Fuller, Li, Yu, & Varadhan                                     [Page 17]
RFC 1338                      Supernetting                     June 1992

   To illustrate the problem alluded to by the "note" in section 4.2,
   consider what happens if RA loses connectivity to C7 (the client
   which is allocated out of RB's space). In a stateful protocol, RA
   will announce to BB that has become
   unreachable. Now, when BB flushes this information out of its routing
   table, any future traffic sent through it for this destination will
   be forwarded to RB (where it will be dropped according to Rule #2) by
   virtue of RB's less specific match  While
   this does not cause an operational problem (C7 is unreachable in any
   case), it does create some extra traffic across "BB" (and may also
   prove confusing to a network manager debugging the outage with
   "traceroute"). A mechanism to cache such unreachability information
   would help here, but is beyond the scope of this document (such a
   mechanism is also not implementable in the near-term).

6.  Transitioning to a long term solution

   This solution does not change the Internet routing and addressing
   architectures.  Hence, transitioning to a more long term solution is
   not affected by the deployment of this plan.

7.  Conclusions

   We are all aware of the growth in routing complexity, and the rapid
   increase in allocation of network numbers.  Given the rate at which
   this growth is being observed, we expect to run out in a few short

   If the inter-domain routing protocol supports carrying network routes
   with associated masks, all of the major concerns demonstrated in this
   paper would be eliminated.

   One of the influential factors which permits maximal exploitation of
   the advantages of this plan is the number of people who agree to use
   it.  It is hoped that having the IAB and the Internet society bless
   this plan would go a long way in the wide deployment, and hence
   benefit of this plan.

   If service providers start charging networks for advertising network
   numbers, this would be a very great incentive to share the address
   space, and hence the associated costs of advertising routes to
   service providers.

8.  Recommendations

   The NIC should begin to hand out large blocks of class-C addresses to
   network service providers.  Each block must fall on bit boundaries
   and should be large enough to serve the provider for two years.

Fuller, Li, Yu, & Varadhan                                     [Page 18]
RFC 1338                      Supernetting                     June 1992

   Further, the NIC should distribute very large blocks to continental
   and national network service organizations to allow additional levels
   of aggregation to take place at the major backbone networks.

   Service providers will further allocate power-of-two blocks of
   class-C addresses from their address space to their subscribers.

   All organizations, including those which are multi-homed, should
   obtain address space from their provider (or one of their providers,
   in the case of the multi-homed).  These blocks should also fall on
   bit boundaries to permit easy route aggregation.

   To allow effective use of this new addressing plan to reduce
   propagated routing information, appropriate IETF WGs will specify the
   modifications needed to Inter-Domain routing protocols.
   Implementation and deployment of these modifications should occur as
   quickly as possible.

9.  Bibliography

   [RFC1247]  Moy, J, "The OSPF Specification  Version 2", January 1991.

10.  Security Considerations

   Security issues are not discussed in this memo.

11.  Authors' Addresses

      Vince Fuller
      Pine Hall 115
      Stanford, CA, 94305-4122
      email: vaf@Stanford.EDU

      Tony Li
      cisco Systems, Inc.
      1525 O'Brien Drive
      Menlo Park, CA 94025
      email: tli@cisco.com

      Jessica (Jie Yun) Yu
      Merit Network, Inc.
      1071 Beal Ave.
      Ann Arbor, MI 48109
      email: jyy@merit.edu

Fuller, Li, Yu, & Varadhan                                     [Page 19]
RFC 1338                      Supernetting                     June 1992

      Kannan Varadhan
      Internet Engineer, OARnet
      1224, Kinnear Road,
      Columbus, OH 43212
      email: kannan@oar.net

Fuller, Li, Yu, & Varadhan                                     [Page 20]