Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.2 (Tek) 9/28/84 based on 9/17/84; site hammer.UUCP Path: utzoo!watmath!clyde!burl!ulysses!allegra!mit-eddie!genrad!decvax!tektronix!orca!hammer!steveg From: steveg@hammer.UUCP (Steve Glaser) Newsgroups: net.lan,net.dcom Subject: Re: ETHERNET on Broadband Message-ID: <1372@hammer.UUCP> Date: Wed, 10-Jul-85 02:21:29 EDT Article-I.D.: hammer.1372 Posted: Wed Jul 10 02:21:29 1985 Date-Received: Fri, 12-Jul-85 03:30:25 EDT References: <2926@decwrl.UUCP> <449@bu-cs.UUCP> <1263@opus.UUCP> <454@bu-cs.UUCP> Reply-To: steveg@hammer.UUCP (Steve Glaser) Organization: Tektronix, Wilsonville OR Lines: 32 Xref: watmath net.lan:911 net.dcom:1095 Summary: >Yes, I would like a black box that just snapped onto an ethernet cable >and broadband tap and magically forwarded to another, remote box with >the same connections, but I don't like the idea of 2 or 3 dedicated >channels to do it. I also wonder exactly what gets forwarded (every >packet? the DEC box certainly doesn't read IP packets.) I guess for now >I am using a couple of Vaxes as this 'black box'. I think Bridge makes a box that connects 2 ethernets over a point to point link (V.35 or whatever at any data rate you care about). It works directly on the Ethernet layer, making no assumptions about what's on top of that (IP, DECNET, XNS, etc.). With a simple RF modem, you can easily get this "point-to-point" link onto a broadband cable. It essentially adapts itself to the ethernet addresses of the stuff on either end and uses that to help it decide what needs forwarding across the link. Given that there are only about 100 hosts per Ethernet the tables aren't that big a deal, especially if you don't need to generate them. I suppose this only really works on protocols like TCP where retransmission times adapt to the network rather than "knowing" what an Ethernet can do (most TCP implementations seem to do this). Also, I don't know what this box does with broadcast packets - I suppose they get forwarded (otherwise ARP etc. wouldn't work). I expect that this solution may break (or at least get bogged down) when folks really start using multicast packets on Ethernet. It certainly will have problems if there is a large speed disparity between 10Mbps and the point-to-point link. Steve Glaser Tektronix Inc. /* usual disclaimer goes here */