From owner-freebsd-hackers Wed Feb 15 12:53:05 1995 Return-Path: hackers-owner Received: (from root@localhost) by freefall.cdrom.com (8.6.9/8.6.6) id MAA04052 for hackers-outgoing; Wed, 15 Feb 1995 12:53:05 -0800 Received: from eel.dataplex.net (EEL.DATAPLEX.NET [199.183.109.245]) by freefall.cdrom.com (8.6.9/8.6.6) with ESMTP id MAA04045 for ; Wed, 15 Feb 1995 12:53:00 -0800 Received: from [199.183.109.242] (cod [199.183.109.242]) by eel.dataplex.net (8.6.9/8.6.9) with SMTP id OAA03113; Wed, 15 Feb 1995 14:52:19 -0600 X-Sender: wacky@shark.dataplex.net Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 15 Feb 1995 14:52:20 -0600 To: Tom Samplonius From: rkw@dataplex.net (Richard Wackerbarth) Subject: Re: Splitting bandwidth Cc: hackers@FreeBSD.org Sender: hackers-owner@FreeBSD.org Precedence: bulk >> In response to Jon's previous message, I concur that a multi-link PPP >> implementation would be preferable. However, given that I am not much of a >> kernel hack, and that there is code that already works (and just needs to be >> ported/hacked) for SLIP, I think I would prefer to tackle the job that looks >> mildly intimidating, rather than the job that looks hopelessly impossible. > > I really don't like the idea of multi-link PPP for load-balancing. >Load-balancing shouldn't be that difficult. It also would nice if it >could be so general to load balance over any number of packet interfaces >(load balance multiple ethernet interfaces?) The problem is that you need to define a MP protocol within the transport protocol. IF you can install the e-net drivers so that all the machines accept the protocol, that would work just fine. One alternate solution is to "tunnel" a MP protocol in IP, just as we already have IP in IP, we could have IP in MP in IP. In any case, you need the protocol layer to negotiate the MP connections. Read RFP 1717. It addresses some of the problems that you might encounter. ---- Richard Wackerbarth rkw@dataplex.net