Network Working Group                                     Charles Hornig
Request for Comments: 894            Symbolics Cambridge Research Center
                                                             April 1984

A Standard for the Transmission of IP Datagrams over Ethernet Networks


Status of this Memo

  This RFC specifies a standard method of encapsulating Internet
  Protocol (IP) [1] datagrams on an Ethernet [2].  This RFC specifies a
  standard protocol for the ARPA-Internet community.

Introduction

  This memo applies to the Ethernet (10-megabit/second, 48-bit
  addresses).  The procedure for transmission of IP datagrams on the
  Experimental Ethernet (3-megabit/second, 8-bit addresses) is
  described in [3].

Frame Format

  IP datagrams are transmitted in standard Ethernet frames.  The type
  field of the Ethernet frame must contain the value hexadecimal 0800.
  The data field contains the IP header followed immediately by the IP
  data.

  The minimum length of the data field of a packet sent over an
  Ethernet is 46 octets.  If necessary, the data field should be padded
  (with octets of zero) to meet the Ethernet minimum frame size.  This
  padding is not part of the IP packet and is not included in the total
  length field of the IP header.

  The minimum length of the data field of a packet sent over an
  Ethernet is 1500 octets, thus the maximum length of an IP datagram
  sent over an Ethernet is 1500 octets.  Implementations are encouraged
  to support full-length packets.  Gateway implementations MUST be
  prepared to accept full-length packets and fragment them if
  necessary.  If a system cannot receive full-length packets, it should
  take steps to discourage others from sending them, such as using the
  TCP Maximum Segment Size option [4].

  Note:  Datagrams on the Ethernet may be longer than the general
  Internet default maximum packet size of 576 octets.  Hosts connected
  to an Ethernet should keep this in mind when sending datagrams to
  hosts not on the same Ethernet.  It may be appropriate to send
  smaller datagrams to avoid unnecessary fragmentation at intermediate
  gateways.  Please see [4] for further information on this point.





Hornig                                                          [Page 1]



RFC 894                                                       April 1984


Address Mappings

  The mapping of 32-bit Internet addresses to 48-bit Ethernet addresses
  can be done several ways.  A static table could be used, or a dynamic
  discovery procedure could be used.

  Static Table

     Each host could be provided with a table of all other hosts on the
     local network with both their Ethernet and Internet addresses.

  Dynamic Discovery

     Mappings between 32-bit Internet addresses and 48-bit Ethernet
     addresses could be accomplished through the Address Resolution
     Protocol (ARP) [5].  Internet addresses are assigned arbitrarily
     on some Internet network.  Each host's implementation must know
     its own Internet address and respond to Ethernet Address
     Resolution packets appropriately.  It should also use ARP to
     translate Internet addresses to Ethernet addresses when needed.

  Broadcast Address

     The broadcast Internet address (the address on that network with a
     host part of all binary ones) should be mapped to the broadcast
     Ethernet address (of all binary ones, FF-FF-FF-FF-FF-FF hex).

  The use of the ARP dynamic discovery procedure is strongly
  recommended.

Trailer Formats

  Some versions of Unix 4.2bsd use a different encapsulation method in
  order to get better network performance with the VAX virtual memory
  architecture.  Consenting systems on the same Ethernet may use this
  format between themselves.

  No host is required to implement it, and no datagrams in this format
  should be sent to any host unless the sender has positive knowledge
  that the recipient will be able to interpret them.  Details of the
  trailer encapsulation may be found in [6].

  (Note:  At the present time Unix 4.2bsd will either always use
  trailers or never use them (per interface), depending on a boot-time
  option.  This is expected to be changed in the future.  Unix 4.2bsd
  also uses a non-standard Internet broadcast address with a host part
  of all zeroes, this may also be changed in the future.)



Hornig                                                          [Page 2]



RFC 894                                                       April 1984


Byte Order

  As described in Appendix B of the Internet Protocol
  specification [1], the IP datagram is transmitted over the Ethernet
  as a series of 8-bit bytes.

References

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

  [2]  "The Ethernet - A Local Area Network", Version 1.0, Digital
  Equipment Corporation, Intel Corporation, Xerox Corporation,
  September 1980.

  [3]  Postel, J., "A Standard for the Transmission of IP Datagrams
  over Experimental Ethernet Networks", RFC-895, USC/Information
  Sciences Institute, April 1984.

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

  [5]  Plummer, D., "An Ethernet Address Resolution Protocol", RFC-826,
  Symbolics Cambridge Research Center, November 1982.

  [6]  Leffler, S., and M. Karels, "Trailer Encapsulations", RFC-893,
  University of California at Berkeley, April 1984.























Hornig                                                          [Page 3]