From owner-freebsd-questions Tue Mar 12 10:16:51 1996 Return-Path: owner-questions Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id KAA02947 for questions-outgoing; Tue, 12 Mar 1996 10:16:51 -0800 (PST) Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.211]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id KAA02940 for ; Tue, 12 Mar 1996 10:16:47 -0800 (PST) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id LAA06081; Tue, 12 Mar 1996 11:11:58 -0700 From: Terry Lambert Message-Id: <199603121811.LAA06081@phaeton.artisoft.com> Subject: Re: Problems To: khetan@chain.iafrica.com (Khetan Gajjar) Date: Tue, 12 Mar 1996 11:11:58 -0700 (MST) Cc: terry@lambert.org, mark@grondar.za, questions@freebsd.org In-Reply-To: from "Khetan Gajjar" at Mar 11, 96 11:19:59 pm X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-questions@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > > Disable TCP extensions. > > Sorry to be SUCH a newbie; where do I disable this ? In inetd, or > services ? /etc/sysconfig; change: tcp_extensions=YES to: tcp_extensions=NO > > With RFC 1323 enabled, the BSD code causes part of the header area > > which is compared to be modified (specifically, a generation number, > > if my memory is correct). > > I actually think it is a routing problem as such. I mean, that is what > I think (I know nothing, so your opinions are very *highly* regarded). > > Do you think it could be due to routing ? Well, I don't think a routing problem would effect the speed; if there were a route problem, the packets just wouldn't get through in one direction or the other or both. For a PPP link, it's not something like "they get through, only slowly" like you would get with an ethernet card with a bogus IRQ setting (only for some cards). > And how do I disable the tcp extensions ? See above. > What advantage/disadvantage are there to them ? The advantage is they enable transactioning. You'd need to read the new Steven's book for a full explanation, but basically, it cuts packet overhead in about half. The problem with PPP is that a seperate internal count structure should be used so that PPP compression can be used. Any changes to the header (including the count structure that is *always* updated by T/TCP), and the compression isn't done on that particular packet. There was a discussion of this on the -hackers list a while back, asking about relative compression statistics when compared to Linux. The Linux numbers were several orders of magnitude better (ie: compression was used more frequently) than BSD because of the compression defeat cause by the header "changes" by T/TCP. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.