RE: Does tcptrace measure RTT for every ACK?

From: Stas Khirman (StasK@narus.com)
Date: 11/17/02

  • Next message: Mark Allman: "Re: Does tcptrace measure RTT for every ACK?" 

    Message-ID: <580E532D9F7A9B4BAE8A130848E0DDA702B1AC8E@franklin.narus.com>
From: Stas Khirman <StasK@narus.com>
Subject: RE: Does tcptrace measure RTT for every ACK? 
Date: Sat, 16 Nov 2002 21:10:09 -0800

    Gentlemen,

    Did you considered to use three-way handshake as a good estimation for RTT?
    Two major advantages to highlight:

    1. As opposite to unpredictable and implementation depended ACK in-host
    delay, three-way handshake responses ( SYN-ACK and ACK) are send by host as
    soon as possible.
    2. Time measured between SYN and ACK is not depend on monitor placement
    (assuming constant latency for almost similar-length SYN and ACK packets).

    Any comments???

    > -----Original Message-----
    > From: Mark Allman [mailto:mallman@grc.nasa.gov]
    > Sent: Friday, November 15, 2002 4:10 AM
    > To: uaca@alumni.uv.es
    > Cc: TCPTRACE Mailing List
    > Subject: Re: Does tcptrace measure RTT for every ACK?
    >
    >
    >
    > > If I'm right RTT meassured values differ their meaning _a
    > lot_ depeding
    > > where the monitorization host is
    > >
    > >
    > > Am I right?
    >
    > Yes. Vantage point matters a ton in all sorts of metrics -- and RTT
    > is one of those, you you have outlined.
    >
    > I think the bottom line is that when you are measuring for some
    > specific purpose your vantage point should be carefully chosen. If,
    > for instance, you are trying to assess packet reordering in the
    > network a sender-side trace does you little good.
    >
    > On the other hand, if you're investigating TCP sender behavior a
    > sender side trace is going to provide a picture that is quite close
    > to what the actual stack sees/does.
    >
    > It's all a tradeoff and it all depends on your goals.
    >
    > allman
    >
    >
    > --
    > Mark Allman -- BBN/NASA GRC -- http://roland.grc.nasa.gov/~mallman/
    > --------------------------------------------------------------
    > --------------
    > To unsubscribe, send a message with body containing
    > "unsubscribe tcptrace" to
    > majordomo@tcptrace.org.
    >
    ----------------------------------------------------------------------------
    To unsubscribe, send a message with body containing "unsubscribe tcptrace" to
    majordomo@tcptrace.org.

    This archive was generated by hypermail 2b30 : 11/17/02 EST