RFC 1476






Network Working Group                                        R. Ullmann
Request for Comments: 1476                 Process Software Corporation
                                                              June 1993


                  RAP: Internet Route Access Protocol

Status of this Memo



   This memo defines an Experimental Protocol for the Internet
   community.  It does not specify an Internet standard.  Discussion and
   suggestions for improvement are requested.  Please refer to the
   current edition of the "IAB Official Protocol Standards" for the
   standardization state and status of this protocol.  Distribution of
   this memo is unlimited.

Abstract



   This RFC describes an open distance vector routing protocol for use
   at all levels of the internet, from isolated LANs to the major
   routers of an international commercial network provider.

Table of Contents



   1.       Introduction  . . . . . . . . . . . . . . . . . . . 2
   1.1       Link-State and Distance-Vector . . . . . . . . . . 3
   1.2       Terminology  . . . . . . . . . . . . . . . . . . . 3
   1.3       Philosophy . . . . . . . . . . . . . . . . . . . . 3
   2.       RAP Protocol  . . . . . . . . . . . . . . . . . . . 4
   2.1       Command Header Format  . . . . . . . . . . . . . . 4
   2.1.1     Length field . . . . . . . . . . . . . . . . . . . 4
   2.1.2     RAP version  . . . . . . . . . . . . . . . . . . . 5
   2.2       RAP Commands . . . . . . . . . . . . . . . . . . . 5
   2.2.1     No operation . . . . . . . . . . . . . . . . . . . 5
   2.2.2     Poll . . . . . . . . . . . . . . . . . . . . . . . 6
   2.2.3     Error  . . . . . . . . . . . . . . . . . . . . . . 7
   2.2.4     Add Route  . . . . . . . . . . . . . . . . . . . . 8
   2.2.5     Purge Route  . . . . . . . . . . . . . . . . . . . 9
   3.       Attributes of Routes  . . . . . . . . . . . . . . . 9
   3.1       Metric and Option Format . . . . . . . . . . . . .10
   3.1.1     Option Class . . . . . . . . . . . . . . . . . .  10
   3.1.2     Type . . . . . . . . . . . . . . . . . . . . . .  10
   3.1.3     Format . . . . . . . . . . . . . . . . . . . . .  11
   3.2       Metrics and Options  . . . . . . . . . . . . . .  11
   3.2.1     Distance . . . . . . . . . . . . . . . . . . . .  12
   3.2.2     Delay  . . . . . . . . . . . . . . . . . . . . .  12
   3.2.3     MTU  . . . . . . . . . . . . . . . . . . . . . .  12
   3.2.4     Bandwidth  . . . . . . . . . . . . . . . . . . .  12



Ullmann                                                         [Page 1]

RFC 1476                          RAP                          June 1993


   3.2.5     Origin . . . . . . . . . . . . . . . . . . . . .  12
   3.2.6     Target . . . . . . . . . . . . . . . . . . . . .  13
   3.2.7     Packet Cost  . . . . . . . . . . . . . . . . . .  13
   3.2.8     Time Cost  . . . . . . . . . . . . . . . . . . .  13
   3.2.9     Source Restriction . . . . . . . . . . . . . . .  14
   3.2.10    Destination Restriction  . . . . . . . . . . . .  14
   3.2.11    Trace  . . . . . . . . . . . . . . . . . . . . .  14
   3.2.12    AUP  . . . . . . . . . . . . . . . . . . . . . .  15
   3.2.13    Public . . . . . . . . . . . . . . . . . . . . .  15
   4.       Procedure   . . . . . . . . . . . . . . . . . . .  15
   4.1       Receiver filtering . . . . . . . . . . . . . . .  16
   4.2       Update of metrics and options  . . . . . . . . .  16
   4.3       Aggregation  . . . . . . . . . . . . . . . . . .  17
   4.4       Active route selection . . . . . . . . . . . . .  17
   4.5       Transmitter filtering  . . . . . . . . . . . . .  17
   4.6       Last resort loop prevention  . . . . . . . . . .  18
   5.       Conclusion  . . . . . . . . . . . . . . . . . . .  18
   6.       Appendix: Real Number Representation  . . . . . .  19
   7.       References  . . . . . . . . . . . . . . . . . . .  20
   8.       Security Considerations . . . . . . . . . . . . .  20
   9.       Author's Address  . . . . . . . . . . . . . . . .  20

1.  Introduction



   RAP is a general protocol for distributing routing information at all
   levels of the Internet, from private LANs to the widest-flung
   international carrier networks.  It does not distinguish between
   "interior" and "exterior" routing (except as restricted by specific
   policy), and therefore is not as restricted nor complex as those
   protocols that have strict level and area definitions in their
   models.

   The protocol encourages the widest possible dissemination of topology
   information, aggregating it only when limits of thrust, bandwidth, or
   administrative policy require it.  Thus RAP permits aggressive use of
   resources to optimize routes where desired, without the restrictions
   inherent in the simplifications of other models.

   While RAP uses IPv7 [RFC1475] addressing internally, it is run over
   both IPv4 and IPv7 networks, and shares routing information between
   them.  A IPv4 router will only be able to activate and propagate
   routes that are defined within the local Administrative Domain (AD),
   loading the version 4 subset of the address into the local IP
   forwarding database.







Ullmann                                                         [Page 2]

RFC 1476                          RAP                          June 1993


1.1  Link-State and Distance-Vector



   Of the two major classes of routing algorithm, link-state and
   distance vector, only distance vector seems to scale from the local
   network (where RIP is existence-proof of its validity) to large scale
   inter-domain policy routing, where the number of links and policies
   exceeds the ability of each router to map the entire network.

   In between, we have OSPF, an open link state (specifically, using
   shortest-path-first analysis of the graph, hence the acronym)
   protocol, with extensive development in intra-area routing.

   Since distance vector has proven useful at both ends of the range, it
   seems reasonable to apply it to the entire range of scales, creating
   a protocol that works automatically on small groups of LANs, but can
   apply fairly arbitrary policy in the largest networks.

   This helps model the real world, where networks are not clearly
   divided into hierarchical domains with identifiable "border" routers,
   but have many links across organizational structure and over back
   fences.

1.2  Terminology



   The RAP protocol propagates routes in the opposite direction to the
   travel of datagrams using the routes.  To avoid confusion explaining
   the routing protocol, several terms are distinguished:

   source          where datagrams come from, the source of the
                   datagrams

   destination     where datagrams go to, the destination of the
                   datagrams

   origin          where routing information originates, the router
                   initially inserting route information into the
                   RAP domain

   target          where routing information goes, the target uses the
                   information to send datagrams

1.3  Philosophy



   Protocols should become simpler as they evolve.







Ullmann                                                         [Page 3]

RFC 1476                          RAP                          June 1993


2.  RAP Protocol



   The RAP protocol operates on TCP port 38, with peers opening a
   symmetric TCP connection between the RAP ports on each system.  Thus
   only one RAP connection exists between any pair of peers.

   RAP is also used on UDP port 38, as a peer discovery method.  Hosts
   (i.e., non-routing systems) may listen to RAP datagrams on this port
   to discover local gateways.  This is in addition to, or in lieu of,
   an Internet Standard gateway discovery protocol, which does not exist
   at this writing.

   The peers then use RAP commands to send each other all routes
   available though the sending peer.  This occurs as a full-duplex
   (i.e., simultaneous) exchange of information, with no acknowledgement
   of individual commands.

   Once the initial exchange has been completed, the peers send only
   updates to routes, new routes, and purge commands to delete routes
   previously offered.

   When the connection is broken, each system purges all routes that had
   been offered by the peer.

2.1  Command Header Format



   Each RAP command starts with a header.  The header contains a length
   field to identify the start of the next packet in the TCP stream, a
   version number, and the code for the command.  On UDP, the length
   field does not appear:  each UDP datagram must contain exactly one
   RAP command and not contain data or padding after the end of the
   command.

     0                   1                   2                   3
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |        length                                                 |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |        RAP version            |       command code            |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

2.1.1  Length field



   The length is a 32 bit unsigned number specifying the offset in bytes
   from the first byte of the length field of this command packet to the
   start of the length field of the next.  The minimum value is 8.
   There is no specific limit to the length of a command packet;
   implementations MUST be able to at least count and skip over a packet



Ullmann                                                         [Page 4]

RFC 1476                          RAP                          June 1993


   that is too large and then MAY send an error indication.

   Each version of the protocol will profile what size should be
   considered the limit for senders, and what (larger) size should be
   considered by receivers to mean that the connection is insane:
   either unsynchronized or worse.

   For version 1 of the protocol, senders MUST NOT send command packets
   greater than 16384 bytes.  Receivers SHOULD consider packets that
   appear to be greater than 162144 bytes in length to be an indication
   of an unrecoverable error.

   Note that these limits probably will not be approached in normal
   operation of version 1 of the protocol; receivers may reasonably
   decline to use routes described by 16K bytes of metrics and policy.
   But even the most memory-restricted implementation MUST be able to
   skip such a command packet.

2.1.2  RAP version



   The version field is a 16 bit unsigned number.  It identifies the
   version of RAP used for that command.  Note that commands with
   different versions may be mixed on the same connection, although the
   usual procedure will be to do the serious protocol (exchanging route
   updates) only at the highest version common to both ends of the
   connection.

   Each side starts the connection by sending a poll command, using the
   highest version supported and continues by using the highest version
   received in any command from the remote.  The response to the poll
   will either be a no-operation packet at that version or an error
   packet at the highest version supported by the remote.

   This document describes version 1 of the RAP protocol.

2.2  RAP Commands



   There five simple RAP commands, described in the following sections.

2.2.1  No operation



   The no operation command serves to reset the poll timer (see next
   section) of the receiver, or (as a side effect) to tell the receiver
   that a particular version is supported.  It never contains option
   specific data and its length is always 8.

   The no operation command is also used in a UDP broadcast to inform
   other systems that the sender is running RAP actively on the network



Ullmann                                                         [Page 5]

RFC 1476                          RAP                          June 1993


   and is both a possible gateway and a candidate peer.  If this command
   is being sent in response to a broadcast poll, it should be sent only
   to the poller.

   A RAP process may send such broadcasts in a startup sequence, or it
   may persist indefinitely to inform other systems coming on line.  If
   it persists, it must not send them more than once every 10 minutes
   (after the initial startup sequence).  If the RAP process sends polls
   as part of its startup, it must not persist in sending them after the
   startup sequence.

   The command code for no-operation is always 0, regardless of RAP
   version.

2.2.2  Poll



   A poll command packet requests that the other side transmit either a
   no-operation packet, or some other packet if sent without delay.
   (i.e., receivers MUST NOT delay a response to a poll by waiting for
   some other packet expected to be queued soon.)

   The poll command code is always 1, regardless of version, and the
   length is always 8.

   Each RAP implementation runs a timer for each connection, to ensure
   that if the other system becomes unreachable, the connection will be
   closed or reset.  The timers run at each end of the connection are
   independent; each system is responsible for sending polls in time to
   reset its own timer.

   The timer MUST be reset (restarted) on the receipt of any RAP packet,
   regardless of whether the version or command code is known.

   In normal operation, if route updates are being sent in both
   directions, polls may not be necessary for long periods of time as
   the timers are continually reset.  When the connection is quiescent,
   both timers will typically get reset as a result of the side with the
   shorter timer doing a poll, and then getting a no-operation in
   response.  RAP implementations MUST NOT be dependent in any way on
   the size or existence of the remote timer.

   An implementation that has access to information from the TCP layer,
   such as the results of TCP layer keepalives, MAY use this instead of
   or in addition to a timer.  However, the use of TCP keepalives is
   discouraged, and this procedure does not ensure that the remote RAP
   process is alive, only that its TCP is accepting data.  Thus a
   failure mode exists that would not exist for active RAP layer polls.




Ullmann                                                         [Page 6]

RFC 1476                          RAP                          June 1993


   The timer MUST be implemented, SHOULD be configurable in at least the
   range 1 to 10 minutes on a per-peer basis, and MAY be infinite
   (disabled) by explicit configuration.

   On UDP, a system (router or non-routing host) may send RAP polls to
   attempt to locate candidate peers or possible gateways.  Such a
   system must not persist in sending polls after its startup sequence,
   except that a system which actually has offered traffic for non-local
   destinations, and has no available gateways, may continue to send
   periodic polls to attempt to acquire a gateway.

2.2.3  Error



   The error packet is used to report an error, whether fatal, serious
   or informational.  It includes a null terminated text description in
   ISO-10646-UTF-1 of the condition, which may be useful to a human
   administrator, and SHOULD be written to a log file.  (The machine is
   not expected to understand the text.)

   Errors are actual failures (in the interpretation of the receiver) to
   use the correct syntax and semantics of the RAP protocol itself, or
   "failure" of the receiver to implement a version of the protocol.
   Other conditions that may require action on the part of the peer
   (such as purging a route) are given their own command codes.

     0                   1                   2                   3
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |        length                                                 |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |        RAP version (1)        |       command code (2)        |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |        error code (0)  [reserved]                             |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |        description                                            |
    +                                                               +
    |                       ...                                     |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   The RAP system receiving an Error packet MUST NOT regard it as fatal,
   and close the connection or discard routes.  If the sending system
   desires the condition to be fatal (unrecoverable), its proper action
   is to close the connection.  This requirement is to prevent the kind
   of failure mode demonstrated by hosts that killed off TCP connections
   on the receipt of ICMP Host-Unreachable notifications, even when the
   condition is transient.  We do not want to discourage the reporting
   of errors, in the way that some implementations avoided sending ICMP
   datagrams to deal with overly sensitive hosts.



Ullmann                                                         [Page 7]

RFC 1476                          RAP                          June 1993


   An error packet MUST NOT be sent in response to something that is (or
   might be) an error packet itself.  Subsequent versions of RAP should
   keep the command code point (2) of the error packet.

2.2.4  Add Route



   The add route command offers a route to the receiving peer.  As noted
   later, it MUST be a route actually loaded into the forwarding
   database of the offering peer at the time the add route is sent.

     0                   1                   2                   3
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |        length                                                 |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |        RAP version (1)        |       command code (3)        |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |        distance               |     (MBZ)     |     mask      |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |        destination network                                    |
    +                                                               +
    |                    ...                                        |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |        route identifier                                       |
    +                                                               +
    |                    ...                                        |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |        metrics and options    ....                            |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   The add route command describes a single offered route, with the
   metrics and other options (such as policies) associated with the
   route.

   Distance is a simple count of the hops to the RAP process (or other
   routing process) that originated the route, incremented every time
   the route is forwarded.  Its initial value may be greater than 1,
   particularily for a route that is administratively configured to
   aggregate routes for a large network or AD.  It may also enter the
   RAP routing domain for the first time with a non-zero distance
   because the route originated in RIP, OSPF, or BGP; if so, the
   distance carried in that protocol is copied into the RAP route.

   The mask is a count of the number of bits of prefix ones in the
   binary representation of the network mask.  Non-contiguous masks are
   not supported directly.  (The destination restriction option may be
   used to give another, non-contiguous, mask; the header mask would
   then describes the number of contiguous ones.)



Ullmann                                                         [Page 8]

RFC 1476                          RAP                          June 1993


   The route identifier is a 64 bit value that the IP forwarding module
   on the sending host can use to rapidly identify the route and the
   next hop for each incoming datagram.  The host receiving the route
   places this identifier into the forward route ID field of the
   datagrams being sent to this host.

   The route ID is also used to uniquely identify the route in the purge
   route operation.

2.2.5  Purge Route



   The purge route command requires that the receiving peer delete a
   route from its database if in use, and requires that it revoke that
   route from any of its peers to whom it has offered the route.  This
   command should preferably be sent before the route is deleted from
   the sending peer's forwarding database, but this is not (cannot be)
   required; it should be sent without delay when the route is removed.

   The command code is 4.  The format is the same as add route without
   any added metrics or options.

   If the route identifier in a purge route command is zero, the command
   requires the deletion of all routes to the destination previously
   offered by this peer.

3.  Attributes of Routes



   There are a rather large number of possible attributes.
   Possibilities include both metrics, and other options describing for
   example policy restrictions and alterations of proximity.  Any
   particular route will usefully carry only a few attributes or none at
   all, particularily on an infrastructure backbone.  A reasonable
   policy for the routers that make up a backbone might be to strip all
   attributes before propagating routes (discarding routes that carry
   attributes with class indications prohibiting this), and then adding
   (for example) an AUP attribute to all routes propagated off of the
   backbone.  A less drastic method would be to simply prefer routes
   with no restrictions, but still propagate a route with restrictions
   if no other is available.

   Most options can occur more than once in a route if there is any
   sensible reason to do so.









Ullmann                                                         [Page 9]

RFC 1476                          RAP                          June 1993


3.1  Metric and Option Format



   Each metric or option for a route begins with a 32 bit header:

     0                   1                   2                   3
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |   length      | C |  format   |           type                |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |        option data                 ...        |   padding     |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                   RAP Option/Metric Header Format

A description of each field:

   length       length of the option or metric
   C            option class, see below
   format       data format
   type         option type identifier
   data         variable length

3.1.1  Option Class



   This field tells implementations what to do with routes containing
   options or metrics they do not understand.  No implementation is
   required to implement (i.e., understand) any given option or metric
   by the RAP specification itself, except for the distance metric in
   the RAP header.

   Classes:

   0        use, propagate, and include this option unmodified
   1        use, propagate, but do not include this option
   2        use this route, but do not propagate it
   3        discard this route

   Note that class 0 is an imperative:  if the route is propagated, the
   option must be included.

   Class and type are entirely orthogonal, different implementations
   might use different classes for the same option or metric.

3.1.2  Type



   The type code identifies the specific option or metric.  The codes
   are part of the option descriptions following.




Ullmann                                                        [Page 10]

RFC 1476                          RAP                          June 1993


   Type 0 indicates a null (no-operation) option.  It should be class
   zero, but an implementation that "understands" the null option may
   decline to propagate it.

   Note that since an implementation may delete an option of class 1 by
   simply setting its type to 0 and forwarding the route description,
   class 1 does not provide any confidentiality of the content of an
   option.

3.1.3  Format



   The format field specifies the format of the data included after the
   option header.  Formats:

   0        none, no data present.
   1        one or more 32-bit signed integers
   2        a character string, null terminated
   3        one or more real numbers
   4        an octet string
   5        one real, followed by a character string

   Format is also orthogonal to type, but a particular type is usually
   only reasonably represented by one format.  This allows decoding of
   all option values for logging and other troubleshooting, even when
   the option type is unknown.  (A new unknown format will still present
   a problem.)

   Format 4, octet string, is to be represented in dotted-decimal byte
   form when printed; it is normally an internet address.

   Format 5 is intended for dimensioned parameters with the character
   string giving the dimension or scale.

3.2  Metrics and Options



   As much as possible, metrics are kept in the base units of bytes and
   seconds, by analogy to the physics systems of MKS (meter-kilogram-
   second) and CGS (centimeter-gram-second) of base units.

   Bytes aren't the real primitive, the bit is.  We are thus using a
   multiple of 8 that isn't part of what one would come to expect from a
   decimal metric system that uses the other prefixes.  However, since K
   (kilo) is often taken to be 1024, and M (mega) to be 1,048,576 (or
   even 1,024,000) we allow this liberty.

   Distance is measured in units also unique to the field.  It is the
   integer number of times that a datagram must be forwarded to reach
   the destination.  (Hop count.)



Ullmann                                                        [Page 11]

RFC 1476                          RAP                          June 1993


3.2.1  Distance



   The Distance metric counts the number of hops on a route; this is
   included in the RAP route command header.

   The initial distance at insertion into the RAP domain by the origin
   of the route MUST be less than or equal to 2z, where z is the number
   of zero bits in the route mask.

   If the origin derives the route from RIP or OSPF, and the distance
   exceeds 2z, the route must not be used.

   When a router originates a route designed to permit aggregation, the
   distance is usefully set to more than 0; this allows simple subset
   aggregation without propagating small distance changes repeatedly as
   the internal diameter of the described network changes.

   For example, for routers designated to announce a default route for
   an AD, with a 24/48 mask, the maximum initial distance is 96.

3.2.2  Delay



   The Delay metric (Type = 2) measures the one-way path delay.  It is
   usually the sum of delays configured for the gateways and interfaces,
   but might also include path segments that are actually measured.

   Format is real (3), with one value.  The units are seconds.

3.2.3  MTU



   The MTU metric (Type = 3) measures the minimum value over the route
   of the Maximum Transmission Unit, i.e., the largest IP datagram that
   can be routed without resulting in fragmentation.

   Format is one integer, measuring the MTU in bytes.

3.2.4  Bandwidth



   The Bandwidth metric (Type = 4) measures the minimum bandwidth of the
   path segments that make up the route.

   Format is one real, representing bandwidth in bytes/second.

3.2.5  Origin



   The origin attribute (type = 5) identifies the router that originally
   inserted the route into the RAP domain.  It is one of the IP
   addresses of the router, format is 4.



Ullmann                                                        [Page 12]

RFC 1476                          RAP                          June 1993


3.2.6  Target



   The target attribute (type = 6) identifies a host or network toward
   which the route should be propagated, regardless of proximity
   filtering that would otherwise occur.  This aids in the establishment
   of tunnels for hosts or subnets "away from home." It can be used to
   force the route to propagate all the way to the home network, or to
   try to propagate a better route to a host that the origin has
   established a connection (e.g., TCP) with.  Note that a router can
   distinguish these two cases during proximity filtering by comparing
   the route described with the host or network identified by the target
   option.

   Format is 4.

3.2.7  Packet Cost



   The packet cost metric (type = 7) measures the actual cost (to
   someone) of sending data over the route.  It is probably either class
   3 or 0.  Format is 5.

   The real number is the cost in currency units/byte.  Tariffs set in
   packets or "segments" should be converted using the nominal (or
   actual path) size.  For example, Sprintnet charges for DAF
   connections within its network are US$1.40/Ksegment thus for segments
   of 64 bytes, the cost is 0.000021875 USD.

   The string is the 3 capital letter ISO code [ISO4217] for the
   currency used.  Funds codes and codes XAU, XBA, XBB, XBC, XBD, and
   XXX are not used.

   If a route already has a packet cost in a different currency
   associated with it, another instance of this option should be added.
   RAP implementations MUST NOT attempt to convert the currency units
   except when actually making a route selection decision.  That is, the
   effects of a currency conversion should never be propagated, except
   for the proper effect of such a selection decision.

3.2.8  Time Cost



   The time cost metric (type = 8) measures the actual cost of holding
   one or more paths in the route open to send data.  It is probably
   either class 3 or 0.  Format is 5.

   The real number is the cost in currency units/second.  For example,
   Sprintnet charges for international connections (to typical
   destinations) are US$10/hour so the cost is 0.002777778 USD.




Ullmann                                                        [Page 13]

RFC 1476                          RAP                          June 1993


   The other notes re codes used and conversions in the previous section
   also apply.

3.2.9  Source Restriction



   A source restriction option (type 9, format 4, class 2 or 3)
   indicates that a route may only be used by datagrams from a
   particular source or set of sources.  The data consists of a network
   or host number, and a mask to qualify it.  If multiple source
   restriction options are included, the restriction is the logical
   union of the sources specified; i.e., any are permitted.

   Source restrictions must be added to routes when the RAP system has
   security filters set in the IP forwarding layer.  This is necessary
   to prevent datagrams from taking "better" routes that end in the
   datagram being silently discarded at the filter.  Note that this
   propagates confidential information about the security configuration,
   but only toward the net authorized to use the route if the RAP
   implementation is careful about where it is propagated.

3.2.10  Destination Restriction



   A destination restriction option (type 10, format 4, class 3) serves
   only to provide a non-contiguous mask, the destination already having
   been specified in the command header.  Data is the destination
   network and mask.

3.2.11  Trace



   Trace (type 11, format 4, class 0) provides an indication that the
   route has propagated through a particular system.  This can be used
   for loop detection, as well as various methods of troubleshooting.
   The data is one internet address, one of the addresses of the system.
   If an arriving route already carries a trace identifying this system,
   and is not an update, it is discarded.  If it is an update, the route
   is purged.

   Trace SHOULD NOT be simply added to every route traversing a system.
   Rather, it should be added (if being used for loop detection) when
   there is a suspicion that a loop has formed.

   When the distance to a destination has increased twice in a row in a
   fairly short period of time, and the number of trace options present
   in the route did not increase as a result of the last update, the RAP
   process should add a trace option identifying itself to the route.
   Effectively, when a loop forms, one router will select itself to be a
   tracer, adding itself and breaking the loop after one more turn.  If
   that fails for some reason, another router will add its trace.  Each



Ullmann                                                        [Page 14]

RFC 1476                          RAP                          June 1993


   router thus depends in the end only on its own trace and will break
   the loop, even if the other routers are using other methods, or
   simply counting-out the route.

3.2.12  AUP



   The AUP (Acceptable Use Policy) option (type 12, format 2, class
   any), tags a route as being useable only according to the policy of a
   network.  This may be used to avoid traversal of the net by (for
   example) commercial traffic, or to prevent un-intentional use of an
   organization's internal net.  (It does not provide a security barrier
   in the sense of forwarding filters, but does provide cooperative
   exchange of information on the useability of a net.)

   The data is a domain name, probably the name of the network, although
   it may be the name of another organization.  E.g., the routers that
   are subject to the NSF AUP might add NSF.NET as the descriptor of
   that policy.

3.2.13  Public



   Public (type 13, format 0, class 2 or 3) marks the route as
   consisting in part of a public broadcast medium.  Examples of a
   public medium are direct radio broadcast or a multi-drop cable in
   which other receivers, not associated with the destination may read
   the traffic.  I.e., a TV cable is a public medium, a LAN within an
   organization is not, even if it can be easily wiretapped.

   This is intended for use by cable TV providers to identify routes
   that should not be used for private communications, in spite of the
   attractively high bandwidth being offered.

4.  Procedure



   Routing information arrives in the RAP process from other peers, from
   (local) static route and interface configuration, and from other
   protocols (e.g., RIP).  The RAP process filters out routes that are
   of no interest (too detailed or too "far away" in the topology) and
   builds an internal database of available routes.

   From this database, it selects routes that are to be active and loads
   them into the IP forwarding database.

   It then advertises those routes to its peers, at a greater distance.







Ullmann                                                        [Page 15]

RFC 1476                          RAP                          June 1993


   -------------------------------------------------------------------

           [incoming routes]
                   |
                   v
           [proximity filtering/aggregation]       [static routes]
                   |                                  |
                   v                                  v
           [route database]  --->  [selected active routes]
                   ^                       |
                   |                       v
           [RIP, etc. routes]      [output filtering]
                                           |
                                           v
                                   [routes advertised]

   -------------------------------------------------------------------

4.1  Receiver filtering



   The first step is to filter out offered routes that are too "far
   away" or too specific.  The filter consists of a maximum distance at
   which a route is considered usable for each possible (contiguous)
   mask.

   Routers that need universal connectivity must either pass through the
   filter all routes regardless of distance (short of "infinity"), and
   use aggregation to reduce them, or have a default route to a router
   that does this.

   The filter may be adjusted dynamically to fit limited resources, but
   if the filter is opened, i.e., made less restrictive, there may be
   routes that have already been offered and discarded that will never
   become available.

4.2  Update of metrics and options



   The process then updates any metrics present on the route to reflect
   the path to the RAP peer.  MTU and bandwidth are minimized, delay and
   cost are added in.  Distance is incremented.  Any unknown options
   cause class-dependent processing:  discarding the option (class 2) or
   route (3), or marking the route as non-propagatable (1).

   Policy options that are known may cause the route to be discarded at
   this stage.






Ullmann                                                        [Page 16]

RFC 1476                          RAP                          June 1993


4.3  Aggregation



   The next step is to aggregate routes that are subsetted by other
   routes through the same peer.  This should not be done automatically
   in every possible case.  The more information that is propagated, the
   more effective the use of forward route identifiers is likely to be,
   particularily in the case of aggregating into a default route.

   In general, a route can be included in an aggregate, and not
   propagated further, if it is through the same peer (next hop) and has
   a smaller distance metric than the containing route.  (Thus datagrams
   will always travel "downhill" as they take more specific routes.)

   The usual case of aggregation is that routes derived from interface
   configurations on the routers from which they originated are subsumed
   into routes offered by routers explicitly configured to route for an
   entire network, area, or AD.  If the larger area becomes partitioned,
   unaggregatable routes will appear (as routes outside the area become
   the shortest distance routes) and traffic will flow around the
   partition.

   Attributes of routes, particularily policy options, may prevent
   aggregation and may result in routes simply being discarded.

   Some information about aggregation also needs to be represented in
   the forwarding database, if the route is made active:  the router
   will need to make a decision as to which forward route identifier to
   use for each datagram arriving on the active route.

4.4  Active route selection



   The router selects those routes to be entered into the IP forwarding
   database and actively used to forward datagrams from the set of
   routes after aggregation, combined with routes derived from other
   protocols such as RIP.  This selection may be made on any combination
   of attributes and options desired by local policy.

4.5  Transmitter filtering



   Finally, the RAP process must decide which routes to offer to its
   peers.  These must be a subset of the active routes, and may in turn
   be a selected subset for each peer.  Arbitrary local policies may be
   used in deciding whether or not to offer any particular route to a
   given peer.

   However, the transmitter must ensure that any datagram filters are
   represented in the offered route, so that the peer (and its peers)
   will not route into a black hole.



Ullmann                                                        [Page 17]

RFC 1476                          RAP                          June 1993


4.6  Last resort loop prevention



   RAP is designed to support many different kinds of routing selection
   algorithms, and allow them to interact to varying extents.  Routes
   can be shared among administrations, and between systems managed with
   more or less sophistication.

   This leaves one absolute requirement:  routing loops must be self-
   healing, regardless of the algorithm used on each host.  There are
   two caveats:

     1.  A loop will not fix itself in the presence of an error that
         continually recurs (thus re-generating the loop)

     2.  The last resort algorithm does not provide rapid breaking of
         loops, only eventual breaking of them even in the absence of
         any intervention by (human) intelligence.

   The algorithm relies on the distance in the RAP route header.  This
   count must be updated (i.e., incremented by one) at each router
   forwarding the route.

   Routers must also impose some limit on the number of hops permitted
   in incoming routes, discarding any routes that exceed the limit.
   This limit is "infinity" in the classic algorithm.  In RIP, infinity
   is 15, much too low for general inter-domain routing.

   In RAP, infinity is defined as 2z + i, where z is the number of zero
   bits in the mask (as described previously) and i is a small number
   which MUST be configurable.

   Note that RAP depends on the last resort algorithm, "counting to
   infinity," much less than predecessors such as RIP.  Routes in the
   RAP domain will usually be purged from the net as the purge route
   command is flooded without the delays typical of periodic broadcast
   algorithms.  Only in some cases will loops form, and they will be
   counted out as fast as the routing processes can exchange the
   information.

5.  Conclusion



   Unlike prior routing protocols, RAP is designed to solve the entire
   problem, from hands-off autoconfiguration of LAN networks, to
   implementing the most complex policies of international carriers.  It
   provides a scaleable solution to carry the Internet forward to a
   future in which essentially all users of data transmission use IP as
   the fabric of their networks.




Ullmann                                                        [Page 18]

RFC 1476                          RAP                          June 1993


6.  Appendix:  Real Number Representation



   Real numbers are represented by a one byte exponent, e, in excess-128
   notation, and a fraction, f, in excess-8388608 notation, with the
   radix point at the right.  (I.e., the "fraction" is actually an
   integer.)

   e is thus in the range 0 to 255, representing exponents (powers of 2)
   in the range 2^-128 to 2^127.

   f is in the range 0 to 16777215, representing numbers from -8388608
   to 8388607

   The value is (f-8338608) x 2^(e-128)

   The real number is not necessarily normalized, but a normalized
   representation will, of course, provide more accuracy for numbers not
   exactly representable.

   Example code, in C:

   #include <math.h>

   typedef struct {
           unsigned e : 8;
           unsigned f : 24;
           } real;

   double a;          /* input value */
   real r;
   double b;          /* output value */
   double d;
   int e32;

   /* convert to real: */

   d = frexp(a, &e32);
   r.e = e32+105;
   r.f = (int)(d*8388608.0) + 8388608;

   /* convert back: */

   b = ldexp((double)r.f - 8388608.0, (int)r.e - 128);








Ullmann                                                        [Page 19]

RFC 1476                          RAP                          June 1993


7.  References



   [ISO3166]   International Organization for Standardization.  Codes
               for the Representation of Names of Countries.  ISO
               3166, ISO, 1988.

   [ISO4217]   International Organization for Standardization.  Codes
               for the representation of currencies and funds.  ISO
               4217, ISO, 1981.

   [RFC791]    Postel, J., "Internet Protocol - DARPA Internet Program
               Protocol Specification", STD 5, RFC 791, DARPA,
               September 1981.

   [RFC1058]   Hedrick, C., "Routing Information Protocol", STD 34,
               RFC 1058, Rutgers University, June 1988.

   [RFC1247]   Moy, J., "OSPF Version 2", RFC 1247, Proteon, Inc.,
               July 1991.

   [RFC1287]   Clark, D., Chapin, L., Cerf, V., Braden, R., and
               R. Hobby, "Towards the Future Internet Architecture",
               RFC 1287, MIT, BBN, CNRI, ISI, UCDavis, December 1991.

   [RFC1338]   Fuller, V., Li, T., Yu, J., and K. Varadhan,
               "Supernetting: an Address Assignment and Aggregation
               Strategy", RFC 1338, BARRNet, cicso, Merit, OARnet,
               June 1992.

   [RFC1475]   Ullmann, R., "TP/IX: The Next Internet", RFC 1475,
               Process Software Corporation, June 1993.

8.  Security Considerations



   Security issues are discussed in sections 3.2.9 and 3.2.12.

9.  Author's Address



   Robert Ullmann
   Process Software Corporation
   959 Concord Street
   Framingham, MA 01701
   USA

   Phone: +1 508 879 6994 x226
   Email: Ariel@Process.COM





Ullmann                                                        [Page 20]