From owner-freebsd-current Thu Feb 12 17:29:07 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id RAA18952 for freebsd-current-outgoing; Thu, 12 Feb 1998 17:29:07 -0800 (PST) (envelope-from owner-freebsd-current@FreeBSD.ORG) Received: (from jmb@localhost) by hub.freebsd.org (8.8.8/8.8.8) id RAA18914; Thu, 12 Feb 1998 17:28:52 -0800 (PST) (envelope-from jmb) From: "Jonathan M. Bresler" Message-Id: <199802130128.RAA18914@hub.freebsd.org> Subject: Re: Heads up: static -ification In-Reply-To: <19980212161814.38690@follo.net> from Eivind Eklund at "Feb 12, 98 04:18:14 pm" To: eivind@yes.no (Eivind Eklund) Date: Thu, 12 Feb 1998 17:28:52 -0800 (PST) Cc: wollman@khavrinen.lcs.mit.edu, eivind@yes.no, current@FreeBSD.ORG X-Mailer: ELM [version 2.4ME+ PL32 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Eivind Eklund wrote: > > That quote is often true for protocols, yes. OTOH, I haven't yet > found a way of implementing advanced protocols that isn't lousy. And > one of the problems with most layered protocol implementations is that > they don't take the layering far enough - they throw in a lot of > indirect calls and non-visible state as their 'layers', instead of > abstracting enough and consider each layer a process. > > At least for operating systems, message-passing systems tend to be > beautifully simple. The main problem is that if you do them in a > memory-protected environment your performance tend to suck. compare nit (netowrk interface tap) and bpf (berkeley packet filter) load due to bpf increases as the number of packets accepted by the filter increases. load due to nit increases as the number of packets on the wire increases. there was a usenix paper about this. jmb To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message