Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 28 Oct 1999 11:07:22 -0400 (EDT)
From:      Garrett Wollman <wollman@khavrinen.lcs.mit.edu>
To:        "Craig Critchley" <craigc@nwlink.com>
Cc:        <net@FreeBSD.ORG>
Subject:   Re: FTP Net Performance 
Message-ID:  <199910281507.LAA91582@khavrinen.lcs.mit.edu>
In-Reply-To: <218001bf211b$791005b0$0201010a@fuzzer.com>
References:  <Pine.BSF.4.10.9910280237560.30200-100000@jade.chc-chimes.com> <218001bf211b$791005b0$0201010a@fuzzer.com>

next in thread | previous in thread | raw e-mail | index | archive | help
<<On Thu, 28 Oct 1999 01:07:26 -0700, "Craig Critchley" <craigc@nwlink.com> said:

> Ah, I've misunderstood the sequence number being ack'd - it's the next byte
> expected by the reciever, eh?  Damn.  I've probably been tearing apart the
> wrong machine.

No, it's the sequence number of the last octet (or control bit)
successfully received *and desegmented*.  The SYN and FIN bits occupy
one unit in sequence space.  TCP's acknowledgements are cumulative, so
ACK(1234) means that all data up to and including sequence number 1234
have been received successfully.  If sequence 1222 is lost, then
regardless of how many later packets are successfully received, TCP
will send ACK(1221) until a segment containing 1222 is received.

In any event, duplicate acknowledgements are a sign of two possible
problems:

1) Packets are getting re-ordered (not likely in your configuration).

2) Packets are getting dropped (which, as you already stated, was
happening about 25% of the time).

TCP will always perform poorly when large numbers of packets are being
dropped, because it must wait for a retransmission timeout (minimum
value 0.5 seconds in -current and 1.5 +/- 0.5 seconds in -stable) and
then slow-start.

-GAWollman

--
Garrett A. Wollman   | O Siem / We are all family / O Siem / We're all the same
wollman@lcs.mit.edu  | O Siem / The fires of freedom 
Opinions not those of| Dance in the burning flame
MIT, LCS, CRS, or NSA|                     - Susan Aglukark and Chad Irschick


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-net" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199910281507.LAA91582>