From owner-freebsd-questions Mon Nov 23 11:30:42 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id LAA05858 for freebsd-questions-outgoing; Mon, 23 Nov 1998 11:30:42 -0800 (PST) (envelope-from owner-freebsd-questions@FreeBSD.ORG) Received: from pp3.shef.ac.uk (pp3.shef.ac.uk [143.167.2.3]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id LAA05847 for ; Mon, 23 Nov 1998 11:30:40 -0800 (PST) (envelope-from nick@dcs.shef.ac.uk) Received: from [143.167.11.162] (helo=dcs.shef.ac.uk) by pp3.shef.ac.uk with esmtp (Exim 2.02 #2) id 0zi197-0002sF-00; Mon, 23 Nov 1998 18:55:37 +0000 Message-ID: <3659AC4E.376CB6A@dcs.shef.ac.uk> Date: Mon, 23 Nov 1998 18:41:18 +0000 From: "Nick A. Fikouras" X-Mailer: Mozilla 4.04 [en] (Win95; I) MIME-Version: 1.0 To: bmah@CA.Sandia.GOV CC: Andrey Tchoritch , "Bruce A. Mah" , "freebsd-questions@FreeBSD.ORG" Subject: Re: Timestamps and nonces in IP packets References: <199811221910.LAA13033@stennis.ca.sandia.gov> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Bruce A. Mah wrote: > If memory serves me right, Andrey Tchoritch wrote: > > On Sun, 22 Nov 1998, Bruce A. Mah wrote: > > > If memory serves me right, "Nick A. Fikouras" wrote: > > > > > > > I've noticed while monitoring TCP communications between FBSD machines > > > > that every IP packet contains timestamps and nonces that take up an > > > > extra 40 octets in the packets. Does anybody know how I can turn off > > > > this setting? > > > > > > Haven't seen anyone write an answer to this yet, so: > > > > > > Edit /etc/rc.conf, and set: > > > > > > tcp_extensions="NO" > > > > > > > > > What is this extensions for? > > The timestamp option in the TCP header allows to sender of a TCP segment to > compute the round-trip time (RTT) for every segment sent and acknowledged. It > places a timestamp in every outgoing segment it sends...the receiver (assuming > it support this option) copies this timestamp back to an appropriate field in > the ACK it sends back for that segment. > Why can't the TCP sequence number be used instead. TCP uses that sequence numbers anyway to acknowledge the safe receipt of data. > Without the use of the timestamp extension, the sending TCP can only estimate > the RTT once per round trip. > I thought it always requires a round trip to measure the Round Trip Time. > IIRC, the other option controlled by tcp_extensions is window scaling. This > allows larger TCP congestion windows than the 64K allows by the original > specification, by multiplying the window sizes by an agreed-upon power of 2. > Useful mostly for high-delay, high-bandwidth paths, such as satellite links. > > Bruce. > Thank you for your response Bruce. 8) nick To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message