Network Working Group                                        J. Garrett
Request for Comments: 1433                       AT&T Bell Laboratories
                                                              J. Hagan
                                            University of Pennsylvania
                                                               J. Wong
                                                AT&T Bell Laboratories
                                                            March 1993


                             Directed ARP

Status of this Memo

  This memo defines an Experimental Protocol for the Internet
  community.  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

  A router with an interface to two IP networks via the same link level
  interface could observe that the two IP networks share the same link
  level network, and could advertise that information to hosts (via
  ICMP Redirects) and routers (via dynamic routing protocols).
  However, a host or router on only one of the IP networks could not
  use that information to communicate directly with hosts and routers
  on the other IP network unless it could resolve IP addresses on the
  "foreign" IP network to their corresponding link level addresses.
  Directed ARP is a dynamic address resolution procedure that enables
  hosts and routers to resolve advertised potential next-hop IP
  addresses on foreign IP networks to their associated link level
  addresses.

Acknowledgments

  The authors are indebted to Joel Halpern of Network Systems
  Corporation and David O'Leary who provided valuable comments and
  insight to the authors, as well as ongoing moral support as the
  presentation of this material evolved through many drafts.  Members
  of the IPLPDN working group also provided valuable comments during
  presentations and through the IPLPDN mailing list.  Chuck Hedrick of
  Rutgers University, Paul Tsuchiya of Bell Communications Research,
  and Doris Tillman of AT&T Bell Laboratories provided early insight as
  well as comments on early drafts.






Garrett, Hagan & Wong                                           [Page 1]

RFC 1433                      Directed ARP                    March 1993


1.  Terminology

  A "link level network" is the upper layer of what is sometimes
  referred to (e.g., OSI parlance) as the "subnetwork", i.e., the
  layers below IP.  The term "link level" is used to avoid potential
  confusion with the term "IP sub-network", and to identify addresses
  (i.e., "link level address") associated with the network used to
  transport IP datagrams.

  From the perspective of a host or router, an IP network is "foreign"
  if the host or router does not have an address on the IP network.

2.  Introduction

  Multiple IP networks may be administered on the same link level
  network (e.g., on a large public data network).  A router with a
  single interface on two IP networks could use existing routing update
  procedures to advertise that the two IP networks shared the same link
  level network.  Cost/performance benefits could be achieved if hosts
  and routers that were not on the same IP network could use that
  advertised information, and exchange packets directly, rather than
  through the dual addressed router.  But a host or router can not send
  packets directly to an IP address without first resolving the IP
  address to its link level address.

  IP address resolution procedures are established independently for
  each IP network.  For example, on an SMDS network [1], address
  resolution may be achieved using the Address Resolution Protocol
  (ARP) [2], with a separate SMDS ARP Request Address (e.g., an SMDS
  Multicast Group Address) associated with each IP network.  A host or
  router that was not configured with the appropriate ARP Request
  Address would have no way to learn the ARP Request Address associated
  with an IP network, and would not send an ARP Request to the
  appropriate ARP Request Address.  On an Ethernet network a host or
  router might guess that an IP address could be resolved by sending an
  ARP Request to the broadcast address.  But if the IP network used a
  different address resolution procedure (e.g., administered address
  resolution tables), the ARP Request might go unanswered.

  Directed ARP is a procedure that enables a router advertising that an
  IP address is on a shared link level network to also aid in resolving
  the IP address to its associated link level address.  By removing
  address resolution constraints, Directed ARP enables dynamic routing
  protocols such as BGP [3] and OSPF [4] to advertise and use routing
  information that leads to next-hop addresses on "foreign" IP
  networks.  In addition, Directed ARP enables routers to advertise
  (via ICMP Redirects) next-hop addresses that are "foreign" to hosts,
  since the hosts can use Directed ARP to resolve the "foreign" next-



Garrett, Hagan & Wong                                           [Page 2]

RFC 1433                      Directed ARP                    March 1993


  hop addresses.

3.  Directed ARP

  Directed ARP uses the normal ARP packet format, and is consistent
  with ARP procedures, as defined in [1] and [2], and with routers and
  hosts that implement those procedures.

3.1  ARP Helper Address

  Hosts and routers maintain routing information, logically organized
  as a routing table.  Each routing table entry associates one or more
  destination IP addresses with a next-hop IP address and a physical
  interface used to forward a packet to the next-hop IP address.  If
  the destination IP address is local (i.e., can be reached without the
  aid of a router), the next-hop IP address is NULL (or a logical
  equivalent, such as the IP address of the associated physical
  interface).  Otherwise, the next-hop IP address is the address of a
  next-hop router.

  A host or router that implements Directed ARP procedures associates
  an ARP Helper Address with each routing table entry.  If the host or
  router has been configured to resolve the next-hop IP address to its
  associated link level address (or to resolve the destination IP
  address, if the next-hop IP address is NULL), the associated ARP
  Helper Address is NULL.  Otherwise, the ARP Helper Address is the IP
  address of the router that provided the routing information
  indicating that the next-hop address was on the same link level
  network as the associated physical interface.  Section 4 provides
  detailed examples of the determination of ARP Helper Addresses by
  dynamic routing procedures.

3.2  Address Resolution Procedures

  To forward an IP packet, a host or router searches its routing table
  for an entry that is the best match based on the destination IP
  address and perhaps other factors (e.g., Type of Service).  The
  selected routing table entry includes the IP address of a next-hop
  router (which may be NULL), the physical interface through which the
  IP packet should be forwarded, an ARP Helper Address (which may be
  NULL), and other information.  The routing function passes the next-
  hop IP address, the physical interface, and the ARP Helper Address to
  the address resolution function.  The address resolution function
  must then resolve the next-hop IP address (or destination IP address
  if the next-hop IP address is NULL) to its associated link level
  address.  The IP packet, the link level address to which the packet
  should be forwarded, and the interface through which the packet
  should be forwarded are then passed to the link level driver



Garrett, Hagan & Wong                                           [Page 3]

RFC 1433                      Directed ARP                    March 1993


  associated with the physical interface.  The link level driver
  encapsulates the IP packet in one or more link level frames (i.e.,
  may do fragmentation) addressed to the associated link level address,
  and forwards the frame(s) through the appropriate physical interface.
  The details of the functions performed are described via C pseudo-
  code below.

  The procedures are organized as two functions, Route() and Resolve(),
  corresponding to routing and address resolution.  In addition, the
  following low level functions are also used:

    Get_Route(IP_Add,Other) returns a pointer to the routing table
     entry with the destination field that best matches IP_Add.  If no
     matching entry is found, NULL is returned.  Other information such
     as Type of Service may be considered in selecting the best route.

    Forward(Packet,Link_Level_Add,Phys_Int) fragments Packet (if
     needed), and encapsulates Packet in one or more Link Level Frames
     addressed to Link_Level_Add, and forwards the frame(s) through
     interface, Phys_Int.

    Look_Up_Add_Res_Table(IP_Add,Phys_Int) returns a pointer to the
     link level address associated with IP_Add in the address
     resolution table associated with interface, Phys_Int.  If IP_Add
     is not found in the address resolution table, NULL is returned.

    Local_Add_Res(IP_Add,Phys_Int) returns a pointer to the Link Level
     address associated with IP_Add, using address resolution
     procedures associated with address, IP_Add, and interface,
     Phys_Int.  If address resolution is unsuccessful, NULL is
     returned.  Note that different address resolution procedures may
     be used for different IP networks.

    Receive_ARP_Response(IP_Add,Phys_Int) returns a pointer to an ARP
     Response received through interface, Phys_Int, that resolves
     IP_Add.  If no ARP response is received, NULL is returned.

    Dest_IP_Add(IP_Packet) returns the IP destination address from
     IP_Packet.

    Next_Hop(Entry) returns the IP address in the next-hop field of
     (routing table) Entry.

    Interface(Entry) returns the physical interface field of (routing
     table) Entry.

    ARP_Helper_Add(Entry) returns the IP address in the ARP Helper
     Address field of (routing table) Entry.



Garrett, Hagan & Wong                                           [Page 4]

RFC 1433                      Directed ARP                    March 1993


    ARP_Request(IP_Add) returns an ARP Request packet with IP_Add as
     the Target IP address.

    Source_Link_Level(ARP_Response) returns the link level address of
     the sender of ARP_Response.




  ROUTE(IP_Packet)
  {
  Entry = Get_Route(Dest_IP_Add(IP_Packet),Other(IP_Packet));
  If (Entry == NULL)  /* No matching entry in routing table */
    Return;  /*  Discard IP_Packet */
  else
    {  /* Resolve next-hop IP address to link level address */
    If (Next_Hop(Entry) != NULL) /* Route packet via next-hop router */
      Next_IP = Next_Hop(Entry);
    else  /* Destination is local */
      Next_IP = Dest_IP_Add(IP_Packet);
    L_L_Add = Resolve(Next_IP,Interface(Entry),ARP_Helper_Add(Entry));
    If (L_L_Add != NULL)
      Forward(IP_Packet,L_L_Add,Interface(Entry));
    else  /* Couldn't resolve next-hop IP address */
      Return;  /* Discard IP_Packet */
    Return;
    }
  }

  Figure 1:  C Pseudo-Code for the Routing function.





















Garrett, Hagan & Wong                                           [Page 5]

RFC 1433                      Directed ARP                    March 1993


  Resolve(IP_Add,Interface,ARP_Help_Add)
  {
  If ((L_L_Add = Look_Up_Add_Res_Table(IP_Add,Interface)) != NULL)
    {   /* Found it in Address Resolution Table */
    Return L_L_Add;
    }
  else
    {
    If (ARP_Help_Add == NULL)
      {  /* Do local Address Resolution Procedure */
      Return Local_Add_Res(IP_Add,Interface);
      }
    else  /* ARP_Help_Add != NULL */
      {
      L_L_ARP_Help_Add = Look_Up_Add_Res_Table(ARP_Help_Add,Interface);
      If (L_L_ARP_Help_Add == NULL)
                             /* Not in Address Resolution Table */
        L_L_ARP_Help_Add = Local_Add_Res(ARP_Help_Add,Interface);
      If (L_L_ARP_Help_Add == NULL)  /* Can't Resolve ARP Helper Add */
        Return NULL;  /*  Address Resolution Failed */
      else
        {  /* ARP for IP_Add */
        Forward(ARP_Request(IP_Add),L_L_ARP_Help_Add,Interface);
        ARP_Resp = Receive_ARP_Response(IP_Add,Interface);
        If (ARP_Resp == NULL) /* No ARP Response (after persistence) */
          Return NULL;  /* Address Resolution Failed */
        else
          Return Source_Link_Level(ARP_Resp);
          }
        }
      }
    }
  }

  Figure 2:  C Pseudo-Code for Address Resolution function.



3.3  Forwarding ARP Requests

  A host that implements Directed ARP procedures uses normal procedures
  to process received ARP Requests.  That is, if the Target IP address
  is the host's address, the host uses normal procedures to respond to
  the ARP Request.  If the Target IP address is not the host's address,
  the host silently discards the ARP Request.

  If the Target IP address of an ARP Request received by a router is
  the router's address, the router uses normal procedures to respond to



Garrett, Hagan & Wong                                           [Page 6]

RFC 1433                      Directed ARP                    March 1993


  the ARP Request.  But if the Target IP address is not the router's
  address, the router may forward the ARP Request back through the same
  interface it was received from, addressed to a Link Level Address
  that corresponds to an ARP Helper Address in the router's routing
  table.  The procedures used to process an ARP Request are described
  via C pseudo-code below.  The function Receive() describes procedures
  followed by hosts and routers, and the function Direct() describes
  additional procedures followed by routers.  In addition, the
  following low level functions are also used:

    Is_Local_IP_Add(IP_Add,Phys_Int) returns TRUE if Phys_Int has been
     assigned IP address, IP_Add.  Otherwise, returns FALSE.

    Do_ARP_Processing(ARP_Request,Interface) processes ARP_Request
     using ARP procedures described in [2].

    I_Am_Router returns TRUE if device is a router and False if device
     is a host.

    Target_IP(ARP_Request) returns the Target IP address from
     ARP_Request.

    Filter(ARP_Request,Phys_Int) returns TRUE if ARP_Request passes
     filtering constraints, and FALSE if filtering constraints are not
     passed.  See section 3.4.

    Forward(Packet,Link_Level_Add,Phys_Int) fragments Packet (if
     needed), and encapsulates Packet in one or more Link Level Frames
     addressed to Link_Level_Add, and forwards the frame(s) through
     interface, Phys_Int.

    Look_Up_Next_Hop_Route_Table(IP_Add) returns a pointer to the
     routing table entry with the next-hop field that matches IP_Add.
     If no matching entry is found, NULL is returned.

    Look_Up_Dest_Route_Table(IP_Add) returns a pointer to the routing
     table entry with the destination field that best matches IP_Add.
     If no matching entry is found, NULL is returned.

    Link_Level_ARP_Req_Add(IP_Add,Phys_Int) returns the link level
     address to which an ARP Request to resolve IP_Add should be
     forwarded.  If ARP is not used to perform local address resolution
     of IP_Add, NULL is returned.

    Local_Add_Res(IP_Add,Phys_Int) returns a pointer to the Link Level
     address associated with IP_Add, using address resolution
     procedures associated with address, IP_Add, and interface,
     Phys_Int.  If address resolution is unsuccessful, NULL is



Garrett, Hagan & Wong                                           [Page 7]

RFC 1433                      Directed ARP                    March 1993


     returned.  Note that different address resolution procedures may
     be used for different IP networks.

    Next_Hop(Entry) returns the IP address in the next-hop field of
     (routing table) Entry.

    Interface(Entry) returns the physical interface field of (routing
     table) Entry.

    ARP_Helper_Add(Entry) returns the IP address in the ARP Helper
     Address field of (routing table) Entry.

    Source_Link_Level(ARP_Request) returns the link level address of
     the sender of ARP_Request.





  Receive(ARP_Request,Interface)
  {
  If (Is_Local_IP_Add(Target_IP(ARP_Request),Interface))
    Do_ARP_Processing(ARP_Request,Interface);
  else  /*  Not my IP Address  */
    If (I_Am_Router)  /*  Hosts don't Direct ARP Requests  */
      If (Filter(ARP_Request,Interface))  /*  Passes Filter Test  */
                                          /*  See Section 3.4  */
        Direct(ARP_Request,Interface);  /*  Directed ARP Procedures  */
  Return;
  }

  Figure 3:  C Pseudo-Code for Receiving ARP Requests.



















Garrett, Hagan & Wong                                           [Page 8]

RFC 1433                      Directed ARP                    March 1993


  Direct(ARP_Request,Phys_Int)
  {
  Entry = Look_Up_Next_Hop_Route_Table(Target_IP(ARP_Request));
  If (Entry == NULL)  /* Target_IP Address is not a next-hop */
    {                 /*  in Routing Table */
    Entry = Look_Up_Dest_Route_Table(Target_IP(ARP_Request));
      If (Entry == NULL)  /* Not a destination either */
        Return;  /* Discard ARP Request */
      else
        If (Next_Hop(Entry) != NULL) /* Not a next-hop and Not local */
          Return;  /* Discard ARP Request */
    }
  If (Interface(Entry) != Phys_Int)
                           /* Must be same physical interface */
    Return;  /* Discard ARP Request */
  If (ARP_Helper_Add(Entry) != NULL)
    {
    L_L_ARP_Helper_Add = Resolve(ARP_Helper_Add(Entry),Phys_Int,NULL);
    If (L_L_ARP_Helper_Add != NULL)
      Forward(ARP_Request,L_L_ARP_Helper_Add,Phys_Int);
        /*  Forward ARP_Request to ARP Helper Address  */
    Return;
    }
  else  /*  Do local address resolution.  */
    {
    L_L_ARP_Req_Add =
               Link_Level_ARP_Req_Add(Target_IP(ARP_Request),Phys_Int);
    If (L_L_ARP_Req_Add != NULL)
      {  /*  Local address resolution procedure is ARP. */
         /*  Forward ARP_Request. */
      Forward(ARP_Request,L_L_ARP_Req_Add,Phys_Int);
      Return;
      }
    else
      {  /*  Local address resolution procedure is not ARP.  */
         /*  Do "published ARP" on behalf of Target IP Address  */
      Target_Link_Level =
                     Local_Add_Res(Target_IP(ARP_Request),Phys_Int);
      If (Target_Link_Level != NULL)  /*  Resolved Address  */
        {
        Forward(ARP_Response,Source_Link_Level(ARP_Request),Phys_Int);
        }
      Return;
      }
    }
  }

  Figure 4:  C Pseudo_Code for Directing ARP Requests.



Garrett, Hagan & Wong                                           [Page 9]

RFC 1433                      Directed ARP                    March 1993


3.4  Filtering Procedures

  A router performing Directed ARP procedures must filter the
  propagation of ARP Request packets to constrain the scope of
  potential "ARP floods" caused by misbehaving routers or hosts, and to
  terminate potential ARP loops that may occur during periods of
  routing protocol instability or as a result of inappropriate manual
  configurations.  Specific procedures to filter the propagation of ARP
  Request packets are beyond the scope of this document.  The following
  procedures are suggested as potential implementations that should be
  sufficient.  Other procedures may be better suited to a particular
  implementation.

  To control the propagation of an "ARP flood", a router performing
  Directed ARP procedures could limit the number of identical ARP
  Requests (i.e., same Source IP address and same Target IP address)
  that it would forward per small time interval (e.g., no more than one
  ARP Request per second).  This is consistent with the procedure
  suggested in [5] to prevent ARP flooding.

  Forwarding of ARP Request packets introduces the possibility of ARP
  loops.  The procedures used to control the scope of potential ARP
  floods may terminate some ARP loops, but additional procedures are
  needed if the time required to traverse a loop is longer than the
  timer used to control ARP floods.  A router could refuse to forward
  more than N identical ARP Requests per T minutes, where N and T are
  administered numbers.  If T and N are chosen so that T/N minutes is
  greater than the maximum time required to traverse a loop, such a
  filter would terminate the loop.  In some cases a host may send more
  than one ARP Request with the same Source IP address,Target IP
  address pair (i.e., N should be greater than 1).  For example, the
  first ARP Request might be lost.  However, once an ARP Response is
  received, a host would normally save the associated information, and
  therefore would not generate an identical ARP Request for a period of
  time on the order of minutes.  Therefore, T may be large enough to
  ensure that T/N is much larger than the time to traverse any loop.

  In some implementations the link level destination address of a frame
  used to transport an ARP Request to a router may be available to the
  router's Directed ARP filtering process.  An important class of
  simple ARP loops will be prevented from starting if a router never
  forwards an ARP Request to the same link level address to which the
  received ARP Request was addressed.  Of course, other procedures such
  as the one described in the paragraph above will stop all loops, and
  are needed, even if filters are implemented that prevent some loops
  from starting.





Garrett, Hagan & Wong                                          [Page 10]

RFC 1433                      Directed ARP                    March 1993


  Host requirements [5] specify that "the packet receive interface
  between the IP layer and the link layer MUST include a flag to
  indicate whether the incoming packet was addressed to a link-level
  broadcast address."  An important class of simple ARP floods can be
  eliminated if routers never forward ARP Requests that were addressed
  to a link-level broadcast address.

4.  Use of Directed ARP by Routing

  The exchange and use of routing information is constrained by
  available address resolution procedures.  A host or router can not
  use a next-hop IP address learned via dynamic routing procedures if
  it is unable to resolve the next-hop IP address to the associated
  link level address.  Without compatible dynamic address resolution
  procedures, a router may not advertise a next-hop address that is not
  on the same IP network as the host or router receiving the
  advertisement.  Directed ARP is a procedure that enables a router
  that advertises routing information to make the routing information
  useful by also providing assistance in resolving the associated
  next-hop IP addresses.

  The following subsections describe the use of Directed ARP to expand
  the scope of ICMP Redirects [6], distance-vector routing protocols
  (e.g., BGP [3]), and link-state routing protocols (e.g., OSPF [4]).

4.1  ICMP Redirect

  If a router forwards a packet to a next-hop address that is on the
  same link level network as the host that originated the packet, the
  router may send an ICMP Redirect to the host.  But a host can not use
  a next-hop address advertised via an ICMP Redirect unless the host
  has a procedure to resolve the advertised next-hop address to its
  associated link level address.  Directed ARP is a procedure that a
  host could use to resolve an advertised next-hop address, even if the
  host does not have an address on the same IP network as the
  advertised next-hop address.

  A host that implements Directed ARP procedures includes an ARP Helper
  Address with each routing table entry.  The ARP Helper Address
  associated with an entry learned via an ICMP Redirect is NULL if the
  associated next-hop address matches a routing table entry with a NULL
  next-hop and a NULL ARP Helper Address (i.e., the host already knows
  how to resolve the next-hop address).  Otherwise, the ARP Helper
  Address is the IP address of the router that sent the ICMP Redirect.
  Note that the router that sent the ICMP Redirect is the current
  next-hop to the advertised destination [5].  Therefore, the host
  should have an entry in its address resolution table for the new ARP
  Helper Address.  If the host is unable to resolve the next-hop IP



Garrett, Hagan & Wong                                          [Page 11]

RFC 1433                      Directed ARP                    March 1993


  address advertised in the ICMP Redirect (e.g., because the associated
  ARP Helper Address is on a foreign IP network; i.e., was learned via
  an old ICMP Redirect, and the address resolution table entry for that
  ARP Helper Address timed out), the host must flush the associated
  routing table entry.  Directed ARP procedures do not recursively use
  Directed ARP to resolve an ARP Helper Address.

  A router that performs Directed ARP procedures might advertise a
  foreign next-hop to a host that does not perform Directed ARP.
  Following existing procedures, the host would silently discard the
  ICMP Redirect.  A router that does not implement Directed ARP should
  not advertise a next-hop on a foreign IP network, as specified by
  existing procedures.  If it did, and the ICMP Redirect was received
  by a host that implemented Directed ARP procedures, the host would
  send an ARP Request for the foreign IP address to the advertising
  router, which would silently discard the ARP Request.  When address
  resolution fails, the host should flush the associated entry from its
  routing table.

  For various reasons a host may ignore an ICMP Redirect and may
  continue to forward packets to the same router that sent the ICMP
  Redirect.  For example, a host that does not implement Directed ARP
  procedures would silently discard an ICMP Redirect advertising a
  next-hop address on a foreign IP network.  Routers should implement
  constraints to control the number of ICMP Redirects sent to hosts.
  For example, a router might limit the number of repeated ICMP
  Redirects sent to a host to no more than N ICMP Redirects per T
  minutes, where N and T are administered values.

4.2  Distance Vector Routing Protocol

  A distance-vector routing protocol provides procedures for a router
  to advertise a destination address (e.g., an IP network), an
  associated next-hop address, and other information (e.g., associated
  metric).  But a router can not use an advertised route unless the
  router has a procedure to resolve the advertised next-hop address to
  its associated link level address.  Directed ARP is a procedure that
  a router could use to resolve an advertised next-hop address, even if
  the router does not have an address on the same IP network as the
  advertised next-hop address.

  The following procedures assume a router only accepts routing updates
  if it knows the IP address of the sender of the update, can resolve
  the IP address of the sender to its associated link level address,
  and has an interface on the same link level network as the sender.

  A router that implements Directed ARP procedures includes an ARP
  Helper Address with each routing table entry.  The ARP Helper Address



Garrett, Hagan & Wong                                          [Page 12]

RFC 1433                      Directed ARP                    March 1993


  associated with an entry learned via a routing protocol update is
  NULL if the associated next-hop address matches a routing table entry
  with a NULL next-hop and NULL ARP Helper Address (i.e., the router
  already knows how to resolve the next-hop address).  Otherwise, the
  ARP Helper Address is the IP address of the router that sent the
  routing update.

  Some distance-vector routing protocols (e.g., BGP [3]) provide syntax
  that would permit a router to advertise an address on a foreign IP
  network as a next-hop.  If a router that implements Directed ARP
  procedures advertises a foreign next-hop IP address to a second
  router that does not implement Directed ARP procedures, the second
  router can not use the advertised foreign next-hop.  Depending on the
  details of the routing protocol implementation, it might be
  appropriate for the first router to also advertise a next-hop that is
  not on a foreign IP network (e.g., itself), perhaps at a higher cost.
  Or, if the routing relationship is an administered connection (e.g.,
  BGP relationships are administered TCP/IP connections), the
  administrative procedure could determine whether foreign next-hop IP
  addresses should be advertised.

  A distance-vector routing protocol could advertise that a destination
  is directly reachable by specifying that the router receiving the
  advertisement is, itself, the next-hop to the destination.  In
  addition, the advertised metric for the route might be zero.  If the
  router did not already have a routing table entry that specified the
  advertised destination was local (i.e., NULL next-hop address), the
  router could add the new route with NULL next-hop, and the IP address
  of the router that sent the update as ARP Helper Address.

4.3  Link State Routing Protocol

  A link-state routing protocol provides procedures for routers to
  identify links to other entities (e.g., other routers and networks),
  determine the state or cost of those links, reliably distribute
  link-state information to other routers in the routing domain, and
  calculate routes based on link-state information received from other
  routers.  A router with an interface to two (or more) IP networks via
  the same link level interface is connected to those IP networks via a
  single link, as described above.  If a router could advertise that it
  used the same link to connect to two (or more) IP networks, and would
  perform Directed ARP procedures, routers on either of the IP networks
  could forward packets directly to hosts and routers on both IP
  networks, using Directed ARP procedures to resolve addresses on the
  foreign IP network.  With Directed ARP, the cost of the direct path
  to the foreign IP network would be less than the cost of the path
  through the router with addresses on both IP networks.




Garrett, Hagan & Wong                                          [Page 13]

RFC 1433                      Directed ARP                    March 1993


  To benefit from Directed ARP procedures, the link-state routing
  protocol must include procedures for a router to advertise
  connectivity to multiple IP networks via the same link, and the
  routing table calculation process must include procedures to
  calculate ARP Helper Addresses and procedures to accurately calculate
  the reduced cost of the path to a foreign IP network reached directly
  via Directed ARP procedures.

  The Shortest Path First algorithm for calculating least cost routes
  is based on work by Dijkstra [7], and was first used in a routing
  protocol by the ARPANET, as described by McQuillan [8].  A router
  constructs its routing table by building a shortest path tree, with
  itself as root.  The process is iterative, starting with no entries
  on the shortest path tree, and the router, itself, as the only entry
  in a list of candidate vertices.  The router then loops on the
  following two steps.

    1.  Remove the entry from the candidate list that is closest to
        root, and add it to the shortest path tree.

    2.  Examine the link state advertisement from the entry added to
        the shortest path tree in step 1.  For each neighbor (i.e.,
        router or IP network to which a link connects)

           - If the neighbor is already on the shortest path tree, do
             nothing.

           - If the neighbor is on the candidate list, recalculate the
             distance from root to the neighbor.  Also recalculate the
             next-hop(s) to the neighbor.

           - If the neighbor is not on the candidate list, calculate
             the distance from root to the neighbor and the next-hop(s)
             from root to the neighbor, and add the neighbor to the
             candidate list.

The process terminates when there are no entries on the candidate list.

To take advantage of Directed ARP procedures, the link-state protocol
must provide procedures to advertise that a router accesses two or more
IP networks via the same link.  In addition, the Shortest Path First
calculation is modified to calculate ARP Helper Addresses and recognize
path cost reductions achieved via Directed ARP.

    1.  If a neighbor under consideration is an IP network, and its
        parent (i.e., the entry added to the shortest path tree in step
        1, above) has advertised that the neighbor is reached via the
        same link as a network that is already on the shortest path



Garrett, Hagan & Wong                                          [Page 14]

RFC 1433                      Directed ARP                    March 1993


        tree, the distance from root and next-hop(s) from root to the
        neighbor are the same as the distance and next-hop(s)
        associated with the network already on the shortest path tree.
        If the ARP Helper Address associated with the network that is
        already on the shortest path tree is not NULL, the neighbor
        also inherits the ARP Helper Address from the network that is
        already on the shortest path tree.

    2.  If the calculated next-hop to the neighbor is not NULL, the
        neighbor inherits the ARP Helper Address from its parent.
        Otherwise, except as described in item 1, the ARP Helper
        Address is the IP address of the next-hop to the neighbor's
        parent.  Note that the next-hop to root is NULL.

  For each router or IP network on the shortest path tree, the Shortest
  Path First algorithm described above must calculate one or more
  next-hops that can be used to access the router or IP network.  A
  router that advertises a link to an IP network must include an IP
  address that can be used by other routers on the IP network when
  using the router as a next-hop.  A router might advertise that it was
  connected to two IP networks via the same link by advertising the
  same next-hop IP address for access from both IP networks.  To
  accommodate the address resolution constraints of routers on both IP
  networks the router might advertise two IP addresses (one from each
  IP network) as next-hop IP addresses for access from both IP
  networks.

5.  Robustness

  Hosts and routers can use Directed ARP to resolve third-party next-
  hop addresses; i.e., next-hop addresses learned from a routing
  protocol peer or current next-hop router.  Undetected failure of a
  third party next-hop can result in a routing "black hole".  To avoid
  "black holes", host requirements [5] specify that a host "...MUST be
  able to detect the failure of a 'next-hop' gateway that is listed in
  its route cache and to choose an alternate gateway."  A host may
  receive feedback from protocol layers above IP (e.g., TCP) that
  indicates the status of a next-hop router, and may use other
  procedures (e.g., ICMP echo) to test the status of a next-hop router.
  But the complexity of routing is borne by routers, whose routing
  information must be consistent with the information known to their
  peers.  Routing protocols such as BGP [3], OSPF [4], and others,
  require that routers must stand behind routing information that they
  advertise.  Routers tag routing information with the IP address of
  the router that advertised the information.  If the information
  becomes invalid, the router that advertised the information must
  advertise that the old information is no longer valid.  If a source
  of routing information becomes unavailable, all information received



Garrett, Hagan & Wong                                          [Page 15]

RFC 1433                      Directed ARP                    March 1993


  from that source must be marked as no longer valid.  The complexity
  of dynamic routing protocols stems from procedures to ensure routers
  either receive routing updates sent by a peer, or are able to
  determine that they did not receive the updates (e.g., because
  connectivity to the peer is no longer available).

  Third-party next-hops can also result in "black holes" if the
  underlying link layer network connectivity is not transitive.  For
  example, SMDS filters [9] could be administered to permit
  communication between the SMDS addresses of router R1 and router R2,
  and between the SMDS addresses of router R2 and router R3, and to
  block communication between the SMDS addresses of router R1 and
  router R3.  Router R2 could advertise router R3 as a next-hop to
  router R1, but SMDS filters would prevent direct communication
  between router R1 and router R3.  Non-symmetric filters might permit
  router R3 to send packets to router R1, but block packets sent by
  router R1 addressed to router R3.

  A host or router could verify link level connectivity with a next-hop
  router by sending an ICMP echo to the link level address of the
  next-hop router.  (Note that the ICMP echo is sent directly to the
  link level address of the next-hop router, and is not routed to the
  IP address of the next-hop router.  If the ICMP echo is routed, it
  may follow a path that does not verify link level connectivity.) This
  test could be performed before adding the associated routing table
  entry, or before the first use of the routing table entry.  Detection
  of subsequent changes in link level connectivity is a dynamic routing
  protocol issue and is beyond the scope of this memo.

References

  [1] Piscitello, D., and J. Lawrence, "The Transmission of IP
      Datagrams over the SMDS Service", RFC 1209, Bell Communications
      Research, March 1991.

  [2] Plummer, D., "An Ethernet Address Resolution Protocol - or -
      Converting Network Protocol Addresses to 48.bit Ethernet Address
      for Transmission on Ethernet Hardware", RFC 826, Symbolics, Inc.,
      November 1982.

  [3] Lougheed, K. and Y. Rekhter, "A Border Gateway Protocol 3 (BGP-
      3)", RFC 1267, cisco Systems and IBM T. J. Watson Research
      Center, October 1991.

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

  [5] Braden, R., editor, "Requirements for Internet Hosts --
      Communication Layers", STD 3, RFC 1122, USC/Information Sciences



Garrett, Hagan & Wong                                          [Page 16]

RFC 1433                      Directed ARP                    March 1993


      Institute, October 1989.

  [6] Postel, J., "Internet Control Message Protocol - DARPA Internet
      Program Protocol Specification", STD 5, RFC 792, USC/Information
      Sciences Institute, September 1981.

  [7] Dijkstra, E. W., "A Note on Two Problems in Connection with
      Graphs", Numerische Mathematik, Vol. 1, pp. 269-271, 1959.

  [8] McQuillan, J. M., I. Richer, and E. C. Rosen, "The New Routing
      Algorithm for the ARPANET", IEEE Transactions on Communications,
      Vol. COM-28, May 1980.

  [9] "Generic System Requirements In Support of Switched Multi-
      megabit Data Service", Technical Reference TR-TSV-000772, Bell
      Communications Research Technical Reference, Issue 1, May 1991.



































Garrett, Hagan & Wong                                          [Page 17]

RFC 1433                      Directed ARP                    March 1993


Security Considerations

  Security issues are not discussed in this memo.

Authors' Addresses

  John Garrett
  AT&T Bell Laboratories
  184 Liberty Corner Road
  Warren, N.J. 07060-0906

  Phone: (908) 580-4719
  EMail: [email protected]


  John Dotts Hagan
  University of Pennsylvania
  Suite 221A
  3401 Walnut Street
  Philadelphia, PA 19104-6228

  Phone: (215) 898-9192
  EMail: [email protected]


  Jeffrey A. Wong
  AT&T Bell Laboratories
  184 Liberty Corner Road
  Warren, N.J. 07060-0906

  Phone: (908) 580-5361
  EMail: [email protected]



















Garrett, Hagan & Wong                                          [Page 18]

You are viewing proxied material from gopher.fnord.one. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.