Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!watmath!clyde!floyd!harpo!seismo!hao!hplabs!sri-unix!v.wales@ucla-locus From: v.wales%ucla-locus@sri-unix.UUCP Newsgroups: net.unix-wizards Subject: Re: Ethernet devices (Bridge CS/1's) Message-ID: <17055@sri-arpa.UUCP> Date: Thu, 1-Mar-84 15:06:52 EST Article-I.D.: sri-arpa.17055 Posted: Thu Mar 1 15:06:52 1984 Date-Received: Sun, 4-Mar-84 03:34:36 EST Lines: 33 From: Rich WalesIn reply to the inquiry by Chris Maloney for information about Ethernet interfaces and peripherals: We have had quite a bit of experience lately with the Bridge CS/1 as a terminal multiplexer. On the whole, they have worked well, but we have had to work closely with Bridge to get various bugs and glitches ironed out of the devices, and we also had to make some accommodations in our system software to take good advantage of them. I am preparing a paper on our experience with the CS/1's which I hope to present at the June USENIX conference, and in order not to deflate said presentation I would rather not go into great detail right now as to how we coped with the CS/1. However, I am willing to say this much now: (1) We had to modify our DZ driver to do CTS/RTS-compatible flow con- trol via the "ring" signal. (I plan to show how we did this in my USENIX paper, so please don't ask me to post the mods now.) (2) Some problems exist in accommodating different terminal speeds. We don't lose any data to slow terminals, owing to the flow control alluded to above, but there is no way to tell the computer what speed the terminal is really set to, so software that does variable padding based on the baud rate tends to lose out. (3) There is currently no way for the computer to get the CS/1 network to tell where a given user's terminal is (sort of like the phone company's not telling you the number of the person calling you). This has obvious system security implications. -- Rich <...!ucbvax!ucla-vax!wales>