From owner-freebsd-hackers Wed Jun 26 09:49:11 1996 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id JAA20116 for hackers-outgoing; Wed, 26 Jun 1996 09:49:11 -0700 (PDT) Received: from premise.CS.Berkeley.EDU (root@premise.CS.Berkeley.EDU [128.32.33.172]) by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id JAA20111 for ; Wed, 26 Jun 1996 09:49:09 -0700 (PDT) Received: from premise.CS.Berkeley.EDU (localhost.Berkeley.EDU [127.0.0.1]) by premise.CS.Berkeley.EDU (8.6.11/8.6.9) with ESMTP id JAA18110; Wed, 26 Jun 1996 09:49:05 -0700 Message-Id: <199606261649.JAA18110@premise.CS.Berkeley.EDU> X-Mailer: exmh version 1.6.7 5/3/96 To: "Ron G. Minnich" cc: hackers@freebsd.org Subject: Re: Frame relay and ATM support: virtual interface per vpi? In-reply-to: Your message of "Wed, 26 Jun 1996 12:06:12 EDT." From: bmah@cs.berkeley.edu (Bruce A. Mah) Reply-to: bmah@cs.berkeley.edu X-Face: g~c`.{#4q0"(V*b#g[i~rXgm*w;:nMfz%_RZLma)UgGN&=j`5vXoU^@n5v4:OO)c["!w)nD/!!~e4Sj7LiT'6*wZ83454H""lb{CC%T37O!!'S$S&D}sem7I[A 2V%N&+ Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Wed, 26 Jun 1996 09:49:04 -0700 Sender: owner-hackers@freebsd.org X-Loop: FreeBSD.org Precedence: bulk "Ron G. Minnich" writes: > > Doesn't it seem that if you do the VIF per VCI model, you're > > restricting yourself to having only a single VC to any given > > destination? In other words, it seems that this model would preclude > > my being able to have multiple VCs to another host. > not at all. you can have lots of vc's per destination. The reason we've > set up to support so many VCs is to support many VCs per host. Again, you > have to move your thinking around a bit from the current networking > model. Once you start doing that, however, the particular way to plug ATM > into the BSD networking architecture is not as obvious as it first seems. > We're even looking at having user-mode striped ATM connections between > applications, and boy does this make for some changes ... Hmmm. So I guess, w.r.t the BSD networking implementation, one would (for example) associate a TCP connection with a route through the appropriate VIF for that connection's VC? Something like that? The way I've been looking at things is more from a router perspective, in which you have to route packets for flows that don't necessarily originate locally (thus the method I had in mind above doesn't work). In this case, you need a packet classifier in the router (my implementation was in the ATM device driver) to help with the multiplexing. It really violates layering because link-layer code is looking at network and transport layer headers. But there's no other way to get all the information needed. Bruce.