From owner-freebsd-net Wed Jun 21 7:46: 9 2000 Delivered-To: freebsd-net@freebsd.org Received: from marsh.owlnet.rice.edu (marsh-49.owlnet.rice.edu [128.42.49.222]) by hub.freebsd.org (Postfix) with ESMTP id 6DDE337B832 for ; Wed, 21 Jun 2000 07:46:06 -0700 (PDT) (envelope-from skoe@owlnet.rice.edu) Received: from jungle.owlnet.rice.edu (jungle.owlnet.rice.edu [128.42.49.103]) by marsh.owlnet.rice.edu (8.9.0/8.9.0) with ESMTP id JAA22154; Wed, 21 Jun 2000 09:46:02 -0500 (CDT) Received: from localhost (skoe@localhost) by jungle.owlnet.rice.edu (8.9.0/8.9.0) with ESMTP id JAA10558; Wed, 21 Jun 2000 09:46:02 -0500 (CDT) X-Authentication-Warning: jungle.owlnet.rice.edu: skoe owned process doing -bs Date: Wed, 21 Jun 2000 09:46:02 -0500 (CDT) From: "Anders Chr. Skoe" To: freebsd-net@FreeBSD.ORG Cc: Julie Schlembach , louie@TransSys.COM Subject: Re: Using timestamp option of ip header (IPOPT_TS) In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-net@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org > Processing the record route with timestamp IP option (and source route) will > cause the packet to generally not be forwarded along the highly optimized > (and in most cases, hardware assisted) path, but to be faulted up to > a CPU to be processed as an exception in a the "slow" path. Julie & I are attempting to implement an edge admission control algorithm (more specifically, the Egress Admission Control developed here at Rice - see http://www.ece.rice.edu/networks/publications.html). This approach only requires us to 'stamp' the packets at the ingress router - the intermediate routers should simply forward them. We read the timestamps at the egress router and perform some calculations (the routers are synchronized using NTP). Our idea was to 'activate' the timestamping only at the ingress, hopefully in a manner such that we only get the 'slow path' at the ingress (and eventually egress) router. We were thinking about using a flag analogous to IPOPT_TS_TSANDADDR (eg. IPOPT_TS_INGRESS) to indicate that the timestamping only be done at this edge router. Do you think this might work? -or will we be forced into the 'slow path' on the intermediate routers as long as there are options present? An alternative would be to use the ip_off and ip_tos bits for the timestamp, but we're particularly wary of using the ip_off bits - after all, they serve a purpose. If anyone knows of an example implementation of ip timestamping (online or printed), we would be most grateful. Once we get the timestamping going, we can start worrying about some of the other issues. > There are other mechanisms available to capture a timestamp of when > the packet was received on the local FreeBSD box if that's the > timestamp you're concerned with. libpcap & tcpdump ;-) Thanks for your help so far. Cheers! Anders To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-net" in the body of the message