From owner-freebsd-hackers Tue Mar 19 04:56:50 1996 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id EAA26452 for hackers-outgoing; Tue, 19 Mar 1996 04:56:50 -0800 (PST) Received: from nixpbe.pdb.sni.de (mail.sni.de [192.109.2.33]) by freefall.freebsd.org (8.7.3/8.7.3) with ESMTP id EAA26446 for ; Tue, 19 Mar 1996 04:56:05 -0800 (PST) Received: (from nerv@localhost) by nixpbe.pdb.sni.de (8.6.12/8.6.12) id NAA08040 for hackers@freebsd.org; Tue, 19 Mar 1996 13:55:33 +0100 Message-Id: <199603191255.NAA08040@nixpbe.pdb.sni.de> Subject: ISDN: "modem" or board? (Was: Microsoft "Get ISDN"?) To: hackers@freebsd.org (FreeBSD hackers), isdn@muc.ditec.de (Distribution List; FreeBSD ISDN) Date: Tue, 19 Mar 96 13:52:53 MET From: Greg Lehey X-Mailer: xmail 2.4 (based on ELM 2.2 PL16) Sender: owner-hackers@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Well, I've been away from the list for a few days, and so much has come in that I don't think it would make sense to respond to each message individually, so here's a summary: 1. Speed of a connection. Some people say "the bottleneck is the B channel, so you can use async instead". Well, yes, assuming your machine isn't doing anything else. To run 2 B channels flat out, you need a 230 kb/s line, which with standard el cheapo 16550As will give you 23000 interrupts per second. This is enough to max out a slow 386. In my experience, it's also enough to cause overruns even on fast machines. By contrast, I run my current ISDN card on an 8 MHz 286, and if it drops anything, it doesn't notice :-) In any case, I get full B channel throughput. 2. Setup time. Could be that I've bitten off more than I can chew here. Of course, the setup time is dependent on the switch you connect to, but it also depends on the way you talk to it. On an ISDN board, the software talks directly to the D channel. On an ISDN "modem", you need first to establish (serial) connection with the "modem", it then needs to interpret your commands and talk to the D channel. This is bound to take longer, but the question remains whether the difference is noticable. If it is, then I consider it also objectionable: I think the 2 second setup time I have to be too long. 3. Somebody said that it's nice to be able to leverage off existing technology. That's a nice way of saying nothing: after all, the ISDN boards are existing technology too (ISA bus interface--not fast, but a whole lot faster than a serial interface). Conceptually, think of them as slow Ethernet boards. 4. Price. I think some people have not read the messages carefully. I just ordered another batch of boards for DM 133 each, about $90. What combination of ISDN "modem" and serial board can you get for that? Other points: Darryl says that an Ascend P50 can peak 42 kB/s. That seems reasonable with builtin compression. How many B channels? Is this rate sustainable? Hellmuth mentions a number of points: - you can do other things than IP over a board. - 99% of ISDN IP traffic over here is raw IP over HDLC. I disagree. I know a number of ISPs who use PPP, probably because they don't know any better. The difference is important, because currently the FreeBSD implementation can't handle PPP (probably a bug somewhere). - He mentions CAPI. For those of you who didn't want to ask, CAPI is a German API for ISDN boards. To the best of my knowledge, the only implementations run on DOS. Since the CAPI is supplied with the board, this is a serious limitation. dennis says "the \"future\" is in high-speed async". It's difficult to form an opinion about that from my perspective here. Certainly there's not much activity in high-speed async over here. But I suppose it's a case of absolute crud driving out the crud. I think that's covered most of the topics. Greg