From: 吕国晗 (lguohan00@mails.tsinghua.edu.cn)
Date: 08/18/04
Message-ID: <292840128.16122@mails.tsinghua.edu.cn> From: "吕国晗" <lguohan00@mails.tsinghua.edu.cn> Date: Wed, 18 Aug 2004 22:42:08 +0800 Subject: Re:tcptrace How to get the RTT _actually_ sampled by TCP?
I don't know how to do it using tcptrace. But in freebsd,
you can use trpt to get SRTT, and RTTVAR of its TCP connections.(on-line).
Or, probably vern paxson's tcpanaly can do similar thing.
See, Mark Allman, Vern Paxson's "On Estimating End-to-End Network Path
Properties", SIGCOMM99.
在您的来信中曾经提到:
>From: Liangping Ma <lma@mail.eecis.udel.edu>
>Reply-To:
>To: tcptrace@tcptrace.org
>Subject: tcptrace How to get the RTT _actually_ sampled by TCP?
>
>Hello,
>
> I measured the RTTs for a TCP connection using tcptrace. I found that the
> number of RTTs is almost equal to (in fact, a little bit less than) the
> number of packets sent. For example, in one TCP session,
>
> pure pkts sent: 7857
> sack pkts sent: 150
>
> The number of RTTs obtained by TCP is 7614. So tcptrace samples round-trip
> time almost once per packet.
>
> BUT, TCP only samples the round-trip time once per RTT (rather than once
> per packet). [see p. 468 Computer Networks, L. Peterson and B. Davie, 2nd
> Ed., 2000].
>
> How do you get the RTTs that are _actually_ sampled by TCP?
>
> Thanks.
>
> Liangping
>
> ----------------------------------------------------------------------------
> 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 2.1.7 : 08/18/04 EDT