RE: CAL TECH's Superfast TCP

From: Harvey Newstrom (mail@HarveyNewstrom.com)
Date: Wed Jun 04 2003 - 19:47:43 MDT

  • Next message: Party of Citizens: "Re: [SeattleRobotics] Humanoids"

    Spudboy100@aol.com wrote,
    > New System Could Speed Up Internet Downloads
    > Wed Jun 4, 3:28 PM ET Add Technology - Reuters to My Yahoo!
    [....]
    > The Fast TCP system, designed by a team of researchers at California Institute
    > of Technology in Pasadena, runs on the same Internet infrastructure currently
    > used but is designed to be much quicker.

    My bogometer just went off. Something is wrong with this article's description. It just doesn't add up. We must be missing
    something.

    > "The sending computer transmits a pack, waits for a signal from the
    > recipient that acknowledges its safe arrival, and then sends the
    > next packet," New Scientist magazine said on Wednesday.

    That's why TCP isn't used for streaming audio or streaming video. We use UDP which doesn't wait for an acknowledgement and doesn't
    have delays. It just shoots out the packets as fast as it can, and the end site tries to catch them as fast as it can, with no
    delays or acknowledgements. We already use this today on the Internet for downloading movies.

    > When the researchers tested 10 Fast TCP systems together it boosted the speed
    > to more than 6,000 times the capacity of the ordinary broadband links.

    This would mean that their previous systems were only utilizing 1/6000th of the broadband connection, less then 1/10th of one
    percent. Surely it couldn't be that bad. What am I missing here?

    A Google Search found <http://netlab.caltech.edu/FAST/> which describes the FAST TCP at CalTech. Their examples only show a
    doubling or tripling of speed. This is wonderfully efficient, but nothing like 6000x mentioned in the article above. I have no
    idea what they were trying to say.

    --
    Harvey Newstrom, CISM, CISSP, IAM, GSEC, IBMCP
    <HarveyNewstrom.com> <Newstaff.com>
    


    This archive was generated by hypermail 2.1.5 : Wed Jun 04 2003 - 20:02:21 MDT