Network Working Group                                            D. Katz
Request for Comments: 1188                                  Merit/NSFNET
Obsoletes:  RFC 1103                                        October 1990


             A Proposed Standard for the Transmission of
                   IP Datagrams over FDDI Networks

Status of this Memo

  This memo defines a method of encapsulating the Internet Protocol
  (IP) datagrams and Address Resolution Protocol (ARP) requests and
  replies on Fiber Distributed Data Interface (FDDI) Networks.  This
  RFC specifies a protocol on the IAB Standards Track for the Internet
  community, and requests discussion and suggestions for improvements.
  Please refer to the current edition of the "IAB Official Protocol
  Standards" for the standardization state and status of this protocol.

  This proposal is the product of the IP over FDDI Working Group of the
  Internet Engineering Task Force (IETF).  Comments on this memo should
  be submitted to the IETF IP over FDDI Working Group Chair.
  Distribution of this memo is unlimited.

Abstract

  This document specifies a method for the use of IP and ARP on FDDI
  networks.  The encapsulation method used is described, as well as
  various media-specific issues.

Acknowledgments

  This memo draws heavily in both concept and text from RFC 1042 [3],
  written by Jon Postel and Joyce K. Reynolds of USC/Information
  Sciences Institute.  The author would also like to acknowledge the
  contributions of the IP Over FDDI Working Group of the IETF, members
  of ANSI ASC X3T9.5, and others in the FDDI community.

Conventions

  The following language conventions are used in the items of
  specification in this document:

     "Must," "Shall," or "Mandatory"--the item is an absolute
     requirement of the specification.

     "Should" or "Recommended"--the item should generally be followed
     for all but exceptional circumstances.




Katz                                                            [Page 1]

RFC 1188              IP and ARP on FDDI Networks           October 1990


     "May" or "Optional"--the item is truly optional and may be
     followed or ignored according to the needs of the implementor.

Introduction

  The goal of this specification is to allow compatible and
  interoperable implementations for transmitting IP datagrams [1] and
  ARP requests and replies [2].

  The Fiber Distributed Data Interface (FDDI) specifications define a
  family of standards for Local Area Networks (LANs) that provides the
  Physical Layer and Media Access Control Sublayer of the Data Link
  Layer as defined by the ISO Open System Interconnection Reference
  Model (ISO/OSI).  Documents are in various stages of progression
  toward International Standardization for Media Access Control (MAC)
  [4], Physical Layer Protocol (PHY) [5], Physical Layer Medium
  Dependent (PMD) [6], and Station Management (SMT) [7].  The family
  of FDDI standards corresponds to the IEEE 802 MAC layer standards
  [8, 9, 10].

  The remainder of the Data Link Service is provided by the IEEE 802.2
  Logical Link Control (LLC) service [11].  The resulting stack of
  services appears as follows:

       +-------------+
       |   IP/ARP    |
       +-------------+
       |  802.2 LLC  |
       +-------------+-----+
       |  FDDI MAC   | F   |
       +-------------+ D S |
       |  FDDI PHY   | D M |
       +-------------+ I T |
       |  FDDI PMD   |     |
       +-------------+-----+

  This memo describes the use of IP and ARP in this environment.  At
  this time, it is not necessary that the use of IP and ARP be
  consistent between FDDI and IEEE 802 networks, but it is the intent
  of this memo not to preclude Data Link Layer interoperability at such
  time as the standards define it.

  The FDDI standards define both single and dual MAC stations.  This
  document describes the use of IP and ARP on single MAC stations
  (single-attach or dual-attach) only.  Operation on dual MAC stations
  will be described in a forthcoming document.





Katz                                                            [Page 2]

RFC 1188              IP and ARP on FDDI Networks           October 1990


Packet Format

  IP datagrams and ARP requests and replies sent on FDDI networks shall
  be encapsulated within the 802.2 LLC and Sub-Network Access Protocol
  (SNAP) [12] data link layers and the FDDI MAC and physical layers.
  The SNAP must be used with an Organization Code indicating that the
  SNAP header contains the EtherType code (as listed in Assigned
  Numbers [13]).

  802.2 LLC Type 1 communication (which must be implemented by all
  conforming 802.2 stations) is used exclusively.  All frames must be
  transmitted in standard 802.2 LLC Type 1 Unnumbered Information
  format, with the DSAP and the SSAP fields of the 802.2 header set to
  the assigned global SAP value for SNAP [11].  The 24-bit Organization
  Code in the SNAP must be zero, and the remaining 16 bits are the
  EtherType from Assigned Numbers [13] (IP = 2048, ARP = 2054).

     ...--------+--------+--------+
                MAC Header        |                           FDDI MAC
     ...--------+--------+--------+

     +--------+--------+--------+
     | DSAP=K1| SSAP=K1| Control|                            802.2 LLC
     +--------+--------+--------+

     +--------+--------+---------+--------+--------+
     |Protocol Id or Org Code =K2|    EtherType    |        802.2 SNAP
     +--------+--------+---------+--------+--------+

     The total length of the LLC Header and the SNAP header is 8
     octets.

     The K1 value is 170 (decimal).

     The K2 value is 0 (zero).

     The control value is 3 (Unnumbered Information).

Address Resolution

  The mapping of 32-bit Internet addresses to 48-bit FDDI addresses
  shall be done via the dynamic discovery procedure of the Address
  Resolution Protocol (ARP) [2].

  Internet addresses are assigned arbitrarily on Internet networks.
  Each host's implementation must know its own Internet address and
  respond to Address Resolution requests appropriately.  It must also
  use ARP to translate Internet addresses to FDDI addresses when



Katz                                                            [Page 3]

RFC 1188              IP and ARP on FDDI Networks           October 1990


  needed.

  The ARP protocol has several fields that parameterize its use in any
  specific context [2].  These fields are:

     hrd   16 - bits     The Hardware Type Code
     pro   16 - bits     The Protocol Type Code
     hln    8 - bits     Octets in each hardware address
     pln    8 - bits     Octets in each protocol address
     op    16 - bits     Operation Code

  The hardware type code assigned for IEEE 802 networks is 6 [13].  The
  hardware type code assigned for Ethernet networks is 1 [13].
  Unfortunately, differing values between Ethernet and IEEE 802
  networks cause interoperability problems in bridged environments.  In
  order to not preclude interoperability with Ethernets in a bridged
  environment, ARP packets shall be transmitted with a hardware type
  code of 1.  Furthermore, ARP packets shall be accepted if received
  with hardware type codes of either 1 or 6.

  The protocol type code for IP is 2048 [13].

  The hardware address length is 6.

  The protocol address length (for IP) is 4.

  The operation code is 1 for request and 2 for reply.

  In order to not preclude interoperability in a bridged environment,
  the hardware addresses in ARP packets (ar$sha, ar$tha) must be
  carried in "canonical" bit order, with the Group bit positioned as
  the low order bit of the first octet.  As FDDI addresses are normally
  expressed with the Group bit in the high order bit position, the
  addresses must be bit-reversed within each octet.

  Although outside the scope of this document, it is recommended that
  MAC addresses be represented in canonical order in all Network Layer
  protocols carried within the information field of an FDDI frame.

Broadcast Address

  The broadcast Internet address (the address on that network with a
  host part of all binary ones) must be mapped to the broadcast FDDI
  address (of all binary ones) (see [14]).

Multicast Support

  A method of supporting IP multicasting is specified in [15].  This



Katz                                                            [Page 4]

RFC 1188              IP and ARP on FDDI Networks           October 1990


  method shall be used in FDDI networks if IP multicasting is to be
  supported.  The use of this method may require the ability to copy
  frames addressed to any one of an arbitrary number of multicast
  (group) addresses.

  An IP multicast address is mapped to an FDDI group address by placing
  the low order 23 bits of the IP address into the low order 23 bits of
  the FDDI group address 01-00-5E-00-00-00 (in "canonical" order).
  [See 13, page 20.]

  For example, the IP multicast address:

        224.255.0.2

  maps to the FDDI group address:

        01-00-5E-7F-00-02

  in which the multicast (group) bit is the low order bit of the first
  octet (canonical order).  When bit-reversed for transmission in the
  destination MAC address field of an FDDI frame (native order), it
  becomes:

        80-00-7A-FE-00-40

  that is, with the multicast (group) bit as the high order bit of the
  first octet, that being the first bit transmitted on the medium.

Trailer Formats

  Some versions of Unix 4.x bsd use a different encapsulation method in
  order to get better network performance with the VAX virtual memory
  architecture.  Hosts directly connected to FDDI networks shall not
  use trailers.

Byte Order

  As described in Appendix B of the Internet Protocol specification
  [1], the IP datagram is transmitted over FDDI networks as a series of
  8-bit bytes.  This byte transmission order has been called "big-
  endian" [16].

MAC Layer Details

  Packet Size

     The FDDI MAC specification [4] defines a maximum frame size of
     9000 symbols (4500 octets) for all frame fields, including four



Katz                                                            [Page 5]

RFC 1188              IP and ARP on FDDI Networks           October 1990


     symbols (two octets) of preamble.  This leaves roughly 4470 octets
     for data after the LLC/SNAP header is taken into account.

     However, in order to allow future extensions to the MAC header and
     frame status fields, it is desirable to reserve additional space
     for MAC overhead.

     Therefore, the MTU of FDDI networks shall be 4352 octets.  This
     provides for 4096 octets of data and 256 octets of headers at the
     network layer and above.  Implementations must not send packets
     larger than the MTU.

     Gateway implementations must be prepared to accept packets as
     large as the MTU and fragment them when necessary.  Gateway
     implementations should be able to accept packets as large as can
     be carried within a maximum size FDDI frame and fragment them.

     Host implementations should be prepared to accept packets as large
     as the MTU; however, hosts must not send datagrams longer than 576
     octets unless they have explicit knowledge that the destination is
     prepared to accept them.  Host implementations may accept packets
     as large as can be carried within a maximum size FDDI frame.  A
     host may communicate its size preference in TCP- based
     applications via the TCP Maximum Segment Size option [17].

     Datagrams on FDDI networks may be longer than the general Internet
     default maximum packet size of 576 octets.  Hosts connected to an
     FDDI network should keep this in mind when sending datagrams to
     hosts that are not on the same local network.  It may be
     appropriate to send smaller datagrams to avoid unnecessary
     fragmentation at intermediate gateways.  Please see [17] for
     further information.

     There is no minimum packet size restriction on FDDI networks.

     In order to not preclude interoperability with Ethernet in a
     bridged environment, FDDI implementations must be prepared to
     receive (and ignore) trailing pad octets.

  Other MAC Layer Issues

     The FDDI MAC specification does not require that 16-bit and 48-
     bit address stations be able to interwork fully.  It does,
     however, require that 16-bit stations have full 48-bit
     functionality, and that both types of stations be able to receive
     frames sent to either size broadcast address.  In order to avoid
     interoperability problems, only 48-bit addresses shall be used
     with IP and ARP.



Katz                                                            [Page 6]

RFC 1188              IP and ARP on FDDI Networks           October 1990


     The FDDI MAC specification defines two classes of LLC frames,
     Asynchronous and Synchronous.  Asynchronous frames are further
     controlled by a priority mechanism and two classes of token,
     Restricted and Unrestricted.  Only the use of Unrestricted tokens
     and Asynchronous frames are required by the standard for FDDI
     interoperability.

     All IP and ARP frames shall be transmitted as Asynchronous LLC
     frames using Unrestricted tokens, and the Priority value is a
     matter of local convention.  Implementations should make the
     priority a tunable parameter for future use.  It is recommended
     that implementations provide for the reception of IP and ARP
     packets in Synchronous frames, as well as Restricted Asynchronous
     frames.

     After packet transmission, FDDI provides Frame Copied (C) and
     Address Recognized (A) indicators.  The use of these indicators is
     a local implementation decision.  Implementations may choose to
     perform link-level retransmission, ARP cache entry invalidation,
     etc., based on the values of these indicators and other
     information.  The semantics of these indicators, especially in the
     presence of bridges, are not well defined as of this writing.
     Implementors are urged to follow the work of ANSI ASC X3T9.5 in
     regard to this issue in order to avoid interoperability problems.

IEEE 802.2 Details

  While not necessary for supporting IP and ARP, all implementations
  must support IEEE 802.2 standard Class I service in order to be
  compliant with 802.2.  Described below is the minimum functionality
  necessary for a conformant station.  Some of the functions are not
  related directly to the support of the SNAP SAP (e.g., responding to
  XID and TEST commands directed to the null or global SAP addresses),
  but are part of a general LLC implementation.  Implementors should
  consult IEEE Std. 802.2 [11] for details.

  802.2 Class I LLC requires the support of Unnumbered Information (UI)
  Commands, eXchange IDentification (XID) Commands and Responses, and
  TEST link (TEST) Commands and Responses.  Stations need not be able
  to transmit XID and TEST commands, but must be able to transmit
  responses.

  Encodings

     Command frames are identified by having the low order bit of the
     SSAP address reset to zero.  Response frames have the low order
     bit of the SSAP address set to one.




Katz                                                            [Page 7]

RFC 1188              IP and ARP on FDDI Networks           October 1990


     The UI command has an LLC control field value of 3.

     The XID command/response has an LLC control field value of 175
     (decimal) if the Poll/Final bit is off or 191 (decimal) if the
     Poll/Final bit is on.

     The TEST command/response has an LLC control field value of 227
     (decimal) if the Poll/Final bit is off or 243 (decimal) if the
     Poll/Final bit is on.

  Elements of Procedure

     UI responses and UI commands with the Poll bit set shall be
     ignored.  UI commands having other than the SNAP SAP in the DSAP
     or SSAP fields shall not be processed as IP or ARP packets.

     When an XID or TEST command is received, an appropriate response
     must be returned.  XID and TEST commands must be responded to only
     if the DSAP is the SNAP SAP (170 decimal), the Null SAP (0
     decimal), or the Global SAP (255 decimal).  XID and TEST commands
     received with other DSAP values must not be responded to unless
     the station supports the addressed service.  Responses to XID and
     TEST frames shall be constructed as follows:

        Destination MAC:  Copied from Source MAC of the command
        Source MAC:  Set to the address of the MAC receiving the
               command
        DSAP:  Copied from SSAP of the command
        SSAP:  Set to 171 decimal (SNAP SAP + Response bit) if the
               DSAP in the command was the SNAP SAP or the Global SAP;
               set to 1 decimal (Null SAP + Response bit) if the DSAP
               in the command was the Null SAP

     When responding to an XID or a TEST command, the value of the
     Final bit in the response must be copied from the value of the
     Poll bit in the command.

     XID response frames must include an 802.2 XID Information field of
     129.1.0 indicating Class I (connectionless) service.

     TEST response frames must echo the information field received in
     the corresponding TEST command frame.









Katz                                                            [Page 8]

RFC 1188              IP and ARP on FDDI Networks           October 1990


Appendix on Numbers

  The IEEE specifies numbers as bit strings with the least significant
  bit first, or bit-wise little-endian order.  The Internet protocols
  are documented in bit-wise big-endian order.  This may cause some
  confusion about the proper values to use for numbers.  Here are the
  conversions for some numbers of interest.

      Number           IEEE        Internet    Internet
                       Binary      Binary      Decimal

      UI               11000000    00000011    3
      SAP for SNAP     01010101    10101010    170
      Global SAP       11111111    11111111    255
      Null SAP         00000000    00000000    0
      XID              11110101    10101111    175
      XID Poll/Final   11111101    10111111    191
      XID Info                                 129.1.0
      TEST             11000111    11100011    227
      TEST Poll/Final  11001111    11110011    243

References

  [1] Postel, J., "Internet Protocol", RFC 791, USC/Information
      Sciences Institute, September 1981.

  [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, MIT, November
      1982.

  [3] Postel, J., and Reynolds, J., "A Standard for the Transmission of
      IP Datagrams over IEEE 802 Networks", RFC 1042, USC/Information
      Sciences Institute, February 1988.

  [4] ISO, "Fiber Distributed Data Interface (FDDI) - Media Access
      Control", ISO 9314-2, 1989.  See also ANSI X3.139-1987.

  [5] ISO, "Fiber Distributed Data Interface (FDDI) - Token Ring
      Physical Layer Protocol", ISO 9314-1, 1989.  See also ANSI
      X3.148-1988.

  [6] ISO, "Fiber Distributed Data Interface (FDDI) - Physical Layer
      Medium Dependent", ISO DIS 9314-3, 1989.  See also ANSI X3.166-
      199x.

  [7] ANSI, "FDDI Station Management", ANSI X3T9.5/84-49 Rev 6.0, 1990.




Katz                                                            [Page 9]

RFC 1188              IP and ARP on FDDI Networks           October 1990


  [8] IEEE, "IEEE Standards for Local Area Networks: Carrier Sense
      Multiple Access with Collision Detection (CSMA/CD) Access Method
      and Physical Layer Specifications", IEEE, New York, New York,
      1985.

  [9] IEEE, "IEEE Standards for Local Area Networks: Token-Passing Bus
      Access Method and Physical Layer Specification", IEEE, New York,
      New York, 1985.

 [10] IEEE, "IEEE Standards for Local Area Networks: Token Ring Access
      Method and Physical Layer Specifications", IEEE, New York, New
      York, 1985.

 [11] IEEE, "IEEE Standards for Local Area Networks: Logical Link
      Control", IEEE, New York, New York, 1985.

 [12] IEEE, "Draft Standard P802.1A--Overview and Architecture", 1989.

 [13] Reynolds, J.K., and J.  Postel, "Assigned Numbers", RFC 1060,
      USC/Information Sciences Institute, March 1990.

 [14] Braden, R., and J.  Postel, "Requirements for Internet Gateways",
      RFC 1009, USC/Information Sciences Institute, June 1987.

 [15] Deering, S., "Host Extensions for IP Multicasting", RFC 1112,
      Stanford University, August 1989.

 [16] Cohen, D., "On Holy Wars and a Plea for Peace", Computer, IEEE,
      October 1981.

 [17] Postel, J., "The TCP Maximum Segment Size Option and Related
      Topics", RFC 879, USC/Information Sciences Institute, November
      1983.

Security Considerations

  Security issues are not discussed in this memo.

Author's Address

  Dave Katz
  Merit/NSFNET
  1075 Beal Ave.
  Ann Arbor, MI  48109

  Phone: (313) 763-4898

  EMail: [email protected]



Katz                                                           [Page 10]

RFC 1188              IP and ARP on FDDI Networks           October 1990





















































Katz                                                           [Page 11]