Re: Life in the Swamps / Testing
Fri, 12 Feb 88 08:58:57 PST


    I love the idea and wish it were mine, but I cannot take credit for
    the flakeway. I stole both the concept and the term from the BBN
    wizards Bill Plummer and Ray Tomlinson. They came up with the first
    flakeway under TOPS-20, during the early TCP/IP days. [I have seen
    the listing of that program, and I think I remember Ray actually
    wrote it; but Bill was responsible for popularizing it in the
    research community at that time [[about 1976-1978]]]. My only
    contribution was to write a flakeway for the UCLA ACP; it ran
    for there many years, reachable via source routing.

    (For those who don't know, "flakeway" is a contraction of "flakey
    gateway", a test gateway implementation set to deliberately
    and randomly reorder, delay, and/or drop packets with some set
    specified frequency distributions. The idea was to test your
    TCP/IP implementation through such a monster).

    From an historical perspective, it is interesting that the
    early work DID include real concern about robustness in the
    presence of long delays and packet losses, and there was active
    testing implementations under such conditions. Then the second-
    generation of implementors came along, and worked entirely in the
    Ethernet environment, promptly ignoring/forgetting the lessons
    already learned. One can speculate that if there were no LAN's
    in the world, we would have discovered (perforce) all the wonderful
    Van Jacobson ideas about 8 years ago.

Bob Braden


This archive was generated by hypermail 2.0b3 on Thu Mar 09 2000 - 14:40:42 GMT