Re: 802 (.2).3 TCP/IP


Dino Farinacci - Control Data - CDCNET TCP/IP Development (dino%CDCCENTR.BITNET@CUNYVM.CUNY.EDU)
Fri, 24 Jun 88 19:17:54 CDT


    This is in response to ssc-vax!bruce@beaver.cs.washington.edu's
    memo <memo <1919@ssc-vax.UUCP>:

    When we (CDC) initially started support of TCP/IP in CDCNET
    we made the decision to support all three of the Ethernet
    header formats. We were informed with the direction of OSI
    standards that 802.2/802.3 was the way to go. Our CDCNET
    systems already supported this header format but we needed to
    include Ethernet 2 and SNAP headers.

    We felt the effort was trivial for incoming frames. Just
    check the ether type field in the 802.3 (or version 2
    header). If the value was greater than 1518 (maximum length
    Ethernet frame), then the header format was Ethernet 2.
    Otherwise it was 802.2/802.3. If 802.2 was present than
    another check is done to determine if a SNAP header follows.
    If so, use the ether type field in the SNAP as one would in
    the Ethernet 2 case. If no SNAP, use the 8-bit SAP value in
    the 802.2 to multiplex to the correct user.

    For outgoing frames, one does not know what header format the
    destination system supports. If it was our own system, it
    didn't matter (we supported all of them). There are two ways
    we allowed this information to be provided. One, staticly
    configure the header format with the IP address/ Ethernet
    address. Or two, let ARP find out. Obviously it was less of
    a burden to the Network Administrator if ARP did it. The
    disadvantage was that ARP broadcasts two ARP requests, each
    with a differnet header format (Ethernet 2 and SNAP - there
    is not an 8-bit 802.2 SAP value assigned for ARP). When the
    reply is received, the IP address/Ethernet address and header
    format is cached.

    We did this over a year and a half ago anticipating that
    workstation vendors would go to the SNAP. However,
    everything I've heard from the TCP/IP interoperability
    conferences indicates no one cares about it and everything
    works fine as it is today.

Dino Farinacci
Control Data Corp.
CDCNET TCP/IP Design and Development
Sunnyvale, CA.



This archive was generated by hypermail 2.0b3 on Thu Mar 09 2000 - 14:42:31 GMT