gore@eecs.nwu.edu (Jacob Gore) (01/24/89)
I heard about dial-up SLIP, but no details. All I know is that Telebit
people are looking forward to it (my words), and that there was a version
developed by BBN and/or CSNET.
Is there any more information available? Is the code available?
Jacob Gore Gore@EECS.NWU.Edu
Northwestern Univ., EECS Dept. {oddjob,gargoyle,att}!nucsrl!gorecraig@NNSC.NSF.NET (Craig Partridge) (01/25/89)
> I heard about dial-up SLIP, but no details. All I know is that Telebit > people are looking forward to it (my words), and that there was a version > developed by BBN and/or CSNET. > Is there any more information available? Is the code available? Jacob: There are several implementations around, in vary forms of stability. To my knowledge only the CSNET one is currently being distributed, and only to CSNET members (the code was developed using CSNET revenue). Other sites with dial-up IP software include the University of Tokyo and BRL. The CSNET software is described in a paper being given next week at the Winter USENIX conference. [L. Lanzillo and C. Partridge, "Implementation of Dial-up IP for UNIX Systems," Proc. 1989 Winter USENIX, San Diego, Calif.]. The key features are: - It drops into binary BSD distributions (e.g. SUN) - It makes connections on demand. When a packet hits your dial-up gateway for a remote site that isn't currently connected, the gateway will dial the phone and establish the link (while the datagram waits). Note that the delay in dialing the phone is long enough that you probably don't want to do more than one such dial-up hop (i.e. dial-up IP works best with star topologies). When the dial-up link has been idle for a few minutes, the line hangs-up. - There are some basic access controls so that the phone link can only be established at times that you agree to incur charges, and only by hosts that you chose to allow to establish the connection. These facilities are optional, and once the link is up, anyone can use it. (We'd like to put in traffic filtering but are still dithering about the best way to do it). I believe the U. Tokyo software is similar [ J. Murai and A. Kato, "Current Status of JUNET," Future Generation Computer Systems, Vol. 4, No. 3, October 1988]. The BRL software (last I heard) requires more active user intervention -- i.e. you dial the phone and then start SLIP over it. Phil Karn also supports some form of dial-up SLIP with his KA9Q package (he's used it to connect to the CSNET dial-up IP gateway). Craig PS: My impression, based on very little testing, is that all these implementations are culturally compatible and that with a tiny bit of work, we could probably make them all interoperate.
leo@SH.CS.NET (Leo Lanzillo) (01/25/89)
>> I heard about dial-up SLIP, but no details. All I know is >> that Telebit people are looking forward to it (my words), and >> that there was a version developed by BBN and/or CSNET. >> Is there any more information available? Is the code available? CSNET offers its members Dialup IP access to relay.cs.net. This service gives sites part-time Internet access. Upon receipt of an IP packet destined for the Internet, it automatically checks to see if the line is up and dials relay.cs.net if necessary. It also allows users to manually bring up the link. We have been using it at CSNET for about 6 months to give IP access to sites which could previously only handle electronic mail. The code is based on Rick Adam's slip as distributed with BSD UNIX. We are monitoring the development of the Point to Point standard and expect to convert to that protocol when it becomes finalized. To CSNET members, we distribute source code which runs under Ultrix, Sun 3.x and BSD UNIX. Sites can use it either to access relay.cs.net, or to set up their own dialup IP network. Contact the CSNET Coordination and Information Center (cic@sh.cs.net) for more details. Leo Lanzillo CSNET Staff.
vixie@decwrl.dec.com (Paul A Vixie) (01/25/89)
Ultrix 3.0 (and perhaps earlier, I don't know) supports dial-up SLIP, in a way. If /usr/new/slattach is run without arguments (as a login shell, e.g.) then it tries to look up the name of the user who runs it in the /etc/sliphosts file. If given an argument, it looks up its argument. The /etc/sliphosts file has entries of the general form: destination gateway netmask speed tty modemtype phonenum logininfo "destination" is the key which /usr/new/slattach looks for (see above); there should also be a name or alias in your DNS or /etc/hosts matching this. "gateway" is the name of your end of the SLIP link, and should also be a name or alias in your DNS or /etc/hosts file. "netmask" is obvious, though it's not _generally_ useful for a point-to-point link in my opinion. "speed" is either the baud rate to use (if it's an outgoing entry) or "any" for incoming entries. "tty" is a /dev/ttyXX name, and no, multiple ttys aren't well-supported. "modemtype" and "phonenum" are used in conjunction with the /etc/acucap file which supports generic modem dialing (also used for uucp and tip). "logininfo" is a "chat script" similar to the one used by UUCP in L.sys. The result is that if you can arrange to execute a specific "slattach" command whenever you want to get connected to your SLIP neighbor, then Ultrix 3.0 has "dial-up SLIP". The CSNET implementation is better in this regard since it is able to make the call when you try to send a packet toward your SLIP neighbor. Both the Ultrix and the CSNET dial-up SLIP implementations are proprietary in one way or another: to get the Ultrix one you need an Ultrix machine (a VAX or a DECstation 3100); to get the CSNET one you need to be a CSNET member. A publically redistributable dial-up SLIP is still very much needed. But a publically-redistributable generic modem dialer is also very much needed -- 4.3bsd uucp and tip both still makes you relink binaries to support a new kind of modem. (What decade is this, again?) -- Paul Vixie Work: vixie@decwrl.dec.com decwrl!vixie +1 415 853 6600 Play: paul@vixie.sf.ca.us vixie!paul +1 415 864 7013