From owner-freebsd-hackers Mon Nov 13 13:37:55 1995 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.6.12/8.6.6) id NAA01101 for hackers-outgoing; Mon, 13 Nov 1995 13:37:55 -0800 Received: from who.cdrom.com (who.cdrom.com [192.216.222.3]) by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id NAA01094 for ; Mon, 13 Nov 1995 13:37:51 -0800 Received: from jhome.DIALix.COM (root@jhome.DIALix.COM [192.203.228.69]) by who.cdrom.com (8.6.12/8.6.11) with ESMTP id NAA01082 for ; Mon, 13 Nov 1995 13:37:03 -0800 Received: (from peter@localhost) by jhome.DIALix.COM (8.6.12/8.6.9) id FAA01129; Tue, 14 Nov 1995 05:28:43 +0800 Date: Tue, 14 Nov 1995 05:28:43 +0800 (WST) From: Peter Wemm To: hackers@freebsd.org Subject: if_mux hack? Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-hackers@freebsd.org Precedence: bulk I saw something in Lin*x that got me thinking.... How does this sound for a crude hack to get multiple slip/ppp/hdlc links working in a load-sharing arrangement: Create a stub if_mux that only accepts packets from IP, and redistributes them to other interfaces that it's been told about below it. Ie: it's if_output routine would take the packets and distribute them to other interface's if_output routines below it. Incoming packets from the lower interfaces would still go direct to IP, but that's no big deal. The basic result is that you would have cheap, simple aggregation of point-to-point links. ie: two 28.8K modems in parallel, you can have your route to the other end pointing to "mux0", and be able to send packets interleaved over both modems for nearly double the throughput. This is something like what Cisco's can do with load balancing.. (They do far more I know, but we talk to a Cisco over ppp this way at the moment, and it's an asymmetric link because we can't distribute packets to two interfaces.) Sound interesting? Is it a "worthy hack"? (especially since Linux has something like it.. :-) Cheers, -Peter