Date: Thu, 14 Nov 2002 07:38:56 +0100 Subject: Re: tcptrace RTT caltulation Message-ID: <20021114063856.GA26780@pusa.informat.uv.es> From: uaca@alumni.uv.es
On Wed, Nov 13, 2002 at 06:36:37PM -0500, Liangping Ma wrote:
> Hi, Ulisses,
>
> The second scenario is not reasonable. When you use tcptrace, you should
> take special care of that. In my previous experiments, I got zero RTTs
> because I put the monitor in a mistaken place (in your case the receiving
> end), as in the second scenario.
The second scenario is reasonable because a TCP connection is full duplex
and a receibe host is also a sender :-)
I think this should be documented somehere, I will post a
README.rtt_measurement to the mantainer when this thread ends
Ulisses
>
> Liangping Ma
>
> On Wed, 13 Nov 2002 uaca@alumni.uv.es wrote:
>
> > On Wed, Nov 13, 2002 at 12:39:57PM -0500, Liangping Ma wrote:
> > > Hi, Ulisses,
> > >
> > > I did TCP RTT measurement before. Based on my memory, the tcptrace
> > > processes tcpdump raw data, which can be collected in a local ethernet.
> > > You are right, that the measurement may not be on one of the end points
> > > of a TCP conneciton. But the delay within a LAN is trivial in the case of
> > > a long-haul TCP connection. Thus the measurement can be reasonably
> > > accurate.
> >
> >
> > Ok, imagine the following two scenarios
> >
> >
> > <Monitorization Host>
> > |
> > |
> > [Host A] >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> [Host B]
> > ((((((((((((((((((((((((((<<<<<<<<<<<<<<<<<<<<<<<<<<<
> >
> >
> >
> > <Monitorization Host>
> > |
> > |
> > [Host A] ))))))))))))))))))))))))))))))))))))))))))))))))))))> [Host B]
> > ((((((((((((((((((((((((((((((((((((((((((((((((((((<
> >
> >
> >
> > If I'm right RTT meassured values differ their meaning _a lot_ depeding
> > where the monitorization host is
> >
> >
> > Am I right?
> >
> >
> > Thanks in advance
> >
> >
> > Ulisses
> >
> >
> > >
> > >
> > > Liangping
> > >
> > > On Wed, 13 Nov 2002 uaca@alumni.uv.es wrote:
> > >
> > > >
> > > > I have read rexmit.c:rtt_ackin() and I belive that RTT meassurement is the
> > > > following:
> > > >
> > > >
> > > > <Monitorization Host>
> > > > |
> > > > |
> > > > [Host A] ))))))))))))))))))))))))))>>>>>>>>>>>>>>>>>>>>>>>>>>> [Host B]
> > > > ((((((((((((((((((((((((((<<<<<<<<<<<<<<<<<<<<<<<<<<<
> > > >
> > > >
> > > > ")" or ">" is the data segment
> > > >
> > > > "(" or "<" is the ACK segment for the corresponding data segment
> > > >
> > > >
> > > > Where the time meassured for RTT is the time that is the sum of ">" and "<"
> > > >
> > > > I suposse here the monitorization is being done in the "middle" of the connection
> > > >
> > > >
> > > > That is, tcptrace's RTT is not like the RTT meassured in the "ping" utility,
> > > > in other words is not a "two full way transmision" meassurement
> > > >
> > > >
> > > >
> > > >
> > > > Thanks in advance
> > > >
> > > >
> > > > Ulisses
> > > >
> > > >
> > > > Debian GNU/Linux: a dream come true
> > > > -----------------------------------------------------------------------------
> > > > "Computers are useless. They can only give answers." Pablo Picasso
> > > >
> > > > ---> Visita http://www.valux.org/ para saber acerca de la <---
> > > > ---> Asociación Valenciana de Usuarios de Linux <---
> > > >
> > > > ----------------------------------------------------------------------------
> > > > To unsubscribe, send a message with body containing "unsubscribe tcptrace" to
> > > > majordomo@tcptrace.org.
> > > >
> > >
> >
> > --
> > Debian GNU/Linux: a dream come true
> > -----------------------------------------------------------------------------
> > "Computers are useless. They can only give answers." Pablo Picasso
> >
> > ---> Visita http://www.valux.org/ para saber acerca de la <---
> > ---> Asociación Valenciana de Usuarios de Linux <---
> >
> > ----------------------------------------------------------------------------
> > To unsubscribe, send a message with body containing "unsubscribe tcptrace" to
> > majordomo@tcptrace.org.
> >
>
--
Debian GNU/Linux: a dream come true
-----------------------------------------------------------------------------
"Computers are useless. They can only give answers." Pablo Picasso
---> Visita http://www.valux.org/ para saber acerca de la <---
---> Asociación Valenciana de Usuarios de Linux <---
----------------------------------------------------------------------------
To unsubscribe, send a message with body containing "unsubscribe tcptrace" to
majordomo@tcptrace.org.
This archive was generated by hypermail 2b30 : 11/14/02 EST