rshuford@well.UUCP (08/30/87)
VMS Users Network Information
Some readers of this medium may be interested to know that a group of
users of the DEC VMS operating system met during the DECUS U.S.
Chapter Spring Symposium (Nashville, 27 April through 1 May A.D. 1987)
with the intention of forming a low-cost store-and-forward network in
which computers running VMS could participate.
The organizer of the VAX SIG Birds-of-a-Feather session at the
symposium was Jamie Hanrahan (of Simpact Associates in San Diego).
Todd Aven of DSS and Kevin Carosso of Hughes Aircraft contributed
technical insight.
VMS systems, of course, have for years had the ability to communicate
through DECnet mail systems and file transfer, but DECnet mail cannot
easily work in a store-and-forward fashion and in any case DECnet
costs money, as a "layered product" from DEC.
The intent of the DECUS working group was to develop free or very
low-cost software that would give functionality similar to the "uucp",
"sendmail", and various news programs available for most versions of
AT&T's Unix operating system.
Connecting this new VMS Users' Network (variously called "VUnet" or
"VMSnet") to the existing UUCP, BITNET, and Internet networks would be
considered desirable by many participants. But because of technical
and administrative considerations, VUnet might develop its own set of
backbone sites, relying on gateways for interchange of information
with the rest of the world. Possibly, some organizations that operate
long-distance DECnet links will allow VUnet to piggyback its traffic
during slack periods. Jamie Hanrahan says that the path of least
resistance for establishing VUnet in the domain world would be to
register each participating organization under the UUCP Zone (this
seems to cost $150 per year), barring unforeseen help in establishing
an independent domain.
The original plan was to write some mail/news human-interface software
and use C-Kermit (with scripting and autodial) as the transport
mechanism. (It was considered too difficult in the short term to
reproduce the "uuicio" protocol of "uucp".) But the BOF decided that
the best means of quickly getting the network on the air is to use an
inexpensive piece of software called PMDF.
PMDF implements the identical transport protocol used by MMDF
("Multichannel Memo Distribution Facility" for 4.3 BSD Unix--PMDF is
the "Pascal Memo Distribution Facility"). PMDF was written by Ira
Winston (at Penn State? with CS-Net support) and is now supported for
VMS by Ned Freed of Harvey Mudd College.
PMDF supports script files, and it is said that the program can be
adapted in 30 minutes to support most new types of modem. (Source
code in Pascal, .OBJ files, and an .EXE file are provided.)
Documentation is supplied on the distribution medium (usually a
9-track BACKUP magtape); when printed, it amounts to about 100 pages.
PMDF uses VMS mail as its user interface; Ned has no plans to
implement any other. (It uses a little-known hook in VMS Mail: if an
address contains a string of the form "XN%something", Mail looks for a
sharable image named after "XN".) In addition to dial-up modem links,
PMDF supports other kinds of communication channels, including Mail-11
(which the Interlink gateway software running under VM/CMS on a IBM
System/370 can do), DECnet, Jnet, and several styles of TCP/IP.
Ned says that PMDF supports all major features of the RFC-822 mail
specification (except for domain literals). Version 2.4, released in
July, contains a mechanism for specifying explicit routes for
addressing replies back to originating sites which are not well known,
or which employ unusual gateways. PMDF will attempt delivery over a
12-day period before giving up.
The CS-Net admininstrators want Harvey Mudd College to be the sole
distribution point for VMS PMDF, so each site must order separately
from there. But PMDF does come with what amounts to a site license.
Ned maintains a mailing-list distribution for a continuing electronically
published circular called Info-PMDF. This emanates from
ipmdf@ymir.bitnet or seismo!ymir.bitnet!ipmdf
One other piece of software that is now being investigated (I haven't
talked with anybody who has yet tried to use it) is from the DECUS
Program Library, item VAX-214: NEWS. The program was submitted by
Geoff Huston of the Australian National University in Canberra City,
and from the catalog description it could be a port of one of the
Unix-based news programs, but this is not explicitly stated. VAX-214
could be ordered through regular DECUS mechanisms; the program is also
available from DECUS on the VAX-LIB-6 collection tape.
Otherwise, Jamie is working on a VMS-specific "news" implementation,
consisting of a database server, import and export programs, and a
user interface. One goal is to permit users on different VAXes but on
the same local network to access a common news database residing on a
single machine. The components will communicate using a modified
version of Brian Kantor and Phil Lapsley's NNTP (Network News Transfer
Protocol).
Jamie Hanrahan would like to be contacted by as many interested VMS
sites as possible (and 4.3 BSD sites that would like to help, if any).
He would like to know such things as
- contact information (telephone, e-mail, physical address)
- likelihood of your site paying for some long-distance phone
calls for outgoing traffic or being polled by somebody else
- does your site have a connection to any part of Internet?
- are you willing to pioneer the use of the software?
- does your organization operate any leased phone lines that
have periods of daily disuse (especially DECnet links)?
Several of the Nashville working group have in the past months
obtained PMDF from Harvey Mudd College and have been experimenting
with its telephone dial-up links. They hope to have several VUnet
sites in full operation by December 1987, the time of the DECUS
Symposium in Anaheim.
See also the August 1987 \Pageswapper/, pages VAX-19 thru VAX-31, in
which Jamie wrote his own account of these proceedings. (This is the
VAX SIG newsletter published as part of the joint DECUS SIG
Newsletters.) Follow-up \Pageswapper/ articles are planned.
Contact information for the people involved:
Jamie Hanrahan: sdcsvax!crash!pnet01!jeh or jeh@pnet01.cts.com
Simpact Associates, 9210 Sky Park Court, San Diego, CA 92123
619/565-1865
Ned Freed: ned@ymir.bitnet or seismo!ymir.bitnet!ned
714/621-8006
Todd Aven: todd@umcincom.bitnet or todd@cincom.umd.edu
Kevin Carosso: kvc@eng.hac.com
VMS users in North America may order the PMDF magtape by sending $50
(U.S.) per site to:
Ned Freed
The PMDF Project
Computing Services
Harvey Mudd College
Claremont, CA 91711
(Make checks payable to "The PMDF Project, Harvey Mudd College".)
My role in this is just to promote VUnet among the DECUS Local Users
Groups in North and South Carolina, but I thought some people might
be interested to hear about it here.
.....Richard S. Shuford
Siecor Corporation RD&E, Hickory, NC 28603-0489
BIX: richard
UUCP: ...hplabs!well!rshufordrshuford@well.UUCP (Richard S. Shuford) (09/04/87)
The following pertains to my earlier posting ("VMS Users Network"--
3856@well.uucp in "comp.os.vms" and "comp.org.decus"). This described
an effort by a working group in the DECUS VAX Special Interest Group to
form a UUCP/Usenet-like store-and-forward network for computer systems
running DEC's VMS operating system. The initial plan to establish a set
of sites communicating over dial-up telephone links using a VMS-based
program called PMDF.
Several readers mailed me questions and comments. This is to clarify some
points that they raised.
I was hoping that my posting would invoke commentary by people who really
are experts on these subjects, but until they speak here are my provisional
answers to several questions regarding "VUnet".
/\/\/\/\/\/\/\/\/\/\/\/\/\/\
ANSWERS TO VUNET QUESTIONS
Q. Doesn't the capability proposed for VUnet exist already in DECnet and
VAXnotes?
A. DECnet does have fully functional electronic mail, but using it to
establish informal store-and-forward links is difficult. And not all
VAX/VMS sites have DECnet.
VAXnotes is a nice conferencing system, but it is not, to my knowl-
edge, compatible with the transmissions of the larger networking
world. And again, neither do all VMS sites have VAXnotes nor is its
purchase easily justified in many situations.
Q. Why invent "yet another protocol"? There are too many already.
A. PMDF is not a new protocol. Many sites in the CS-Net have been using
it, or MMDF, for years.
Q. Why have a completely independent networking system for VMS sites?
A. It is not intended that VUnet should be independent. The desire to
connect with the existing system of global networks has been the
driving force during the entire project.
Q. Why not just use a uucp program running under VMS?
A. No low-cost implementations of uucp now exist that can be run under
VMS. (There do exist some commercial VMS enhancement products with
uucp, notably DEC/Shell and The Wollangong Group's Eunice, but these
are relatively costly.) The working-group members want to keep the
price of participation low.
Of the two attempts to reverse-engineer uucp with public-domain
software (uuslave and dcp), it is not known that either yet runs
under VMS with full functionality. If anyone succeeds in making
one of them run well, the working group would want to hear about
it immediately, but even so, a PMDF-based network could just add
the "uuicio" protocol as another "channel".
Q. Likewise, why not just run the Usenet news software that already
exists?
A. I don't know if the newly available DECUS Library program VAX-214
NEWS is an exception, but as far as I'm aware, the working-group
members have not yet found any version of a Unix news program that
will execute under VMS without their either modifying source code
extensively or somehow stretching the terms of a Unix license. (It's
the VMS sites that don't have access to Unix that most need VUnet.)
Anyone volunteering to help write or obtain a news program would be
gladly thanked by the working group.
Q. Shouldn't the new network be compatible with the TCP/IP protocols?
A. If you have a TCP/IP communication link, PMDF can use it as one
of its "channels".
Q. If there is no uucp capability, how can a VUnet site become connected
with the dial-up UUCP network?
A. I'm told that all Unix sites running 4.3 BSD also possess the MMDF
program. PMDF was designed to talk to MMDF.
Versions of PMDF also exist for systems other than DEC VAXes; for in-
stance, a version called HPMDF runs on Hewlett-Packard 3000 machines.
Q. How does VUnet relate to a DECUS Library program called VAXnet?
A. No connection. VAXnet is a program that lets you dial out from your
VAX through a modem. It's like a fancy version of "SET HOST/DTE".
Q. What is DECUS?
A. Digital Equipment Computer Users Society
United States Chapter
219 Boston Post Road, BP02
Marlboro, MA 01752-1850
617/480-3418
Q. Why hasn't this been done already?
A. I wish I knew.
/\/\/\/\/\/\/\/\/\/\/\/\/\/\
The chief organizer of the VMS Users Network is
Jamie Hanrahan
Simpact Associates
9210 Sky Park Court
San Diego, CA 92123
619/565-1865
{sdcsvax,akgua,hplabs,hp-sdd,nosc}!crash!pnet01!jeh
jeh@pnet01.cts.com
pnet01!jeh@crash.cts.com
/\/\/\/\/\/\/\/\/\/\/\/\/\/\
.....Richard S. Shuford
Siecor Corporation
489 Siecor Park, M/S-RD
Hickory, NC 28603-0489
BIX: richard
UUCP: {hplabs,ptsfa}!well!rshuforddboyes@uoregon.UUCP (David Boyes) (09/05/87)
In article <3887@well.UUCP> rshuford@well.UUCP (Richard S. Shuford) writes: > >A. No low-cost implementations of uucp now exist that can be run under > VMS. (There do exist some commercial VMS enhancement products with > > Of the two attempts to reverse-engineer uucp with public-domain > software (uuslave and dcp), it is not known that either yet runs > under VMS with full functionality. If anyone succeeds in making > one of them run well, the working group would want to hear about > it immediately, but even so, a PMDF-based network could just add > the "uuicio" protocol as another "channel". DCP works just fine under VMS, as well as a bunch of other machine types. Did you get the latest version that was posted to comp.sources.misc? The person that brought up the point of 'why invent another protocol?' is absolutely right -- tools exist, so let's not reinvent the wheel, ok? > >Q. Shouldn't the new network be compatible with the TCP/IP protocols? > SLIP (Serial Line Internet Protocol) is NOT that difficult to implement. The less standard this thing is, the less people are going to want to mess with it. It's just one more hassle to learn or know about. If you're going to do this, make it INVISIBLE and STANDARD. No excuses -- do it right the first time. >A. I'm told that all Unix sites running 4.3 BSD also possess the MMDF > program. PMDF was designed to talk to MMDF. Not true. -- David Boyes ARPA: 556%OREGON1.BITNET@WISCVM.WISC.EDU Systems Division BITNET: 556@OREGON1 University of Oregon Computing Center UUCP: dboyes@uoregon.UUCP
david@ms.uky.edu (David Herron -- Resident E-mail Hack) (09/06/87)
In article <601@uoregon.UUCP> dboyes@drizzle.UUCP (David Boyes) writes: >In article <3887@well.UUCP> rshuford@well.UUCP (Richard S. Shuford) writes: >>A. No low-cost implementations of uucp now exist that can be run under >> VMS. (There do exist some commercial VMS enhancement products with >> ... >DCP works just fine under VMS, as well as a bunch of other machine >types. Did you get the latest version that was posted to >comp.sources.misc? The person that brought up the point of 'why invent >another protocol?' is absolutely right -- tools exist, so let's not >reinvent the wheel, ok? That's nice ... >>Q. Shouldn't the new network be compatible with the TCP/IP protocols? > >SLIP (Serial Line Internet Protocol) is NOT that difficult to implement. >The less standard this thing is, the less people are going to want to >mess with it. It's just one more hassle to learn or know about. If >you're going to do this, make it INVISIBLE and STANDARD. No excuses -- >do it right the first time. SLIP is only the beginning ... you've gotta implement the rest of TCP/IP which ain't no trivial task! Two possibilities here are the CMU TCP/IP for VMS which has the advantage of being free. (but no support (of course)). The other possibility is Phil Karn's TCP/IP which will soon have support to run under a Unix with one process doing IP and a subroutine package doing TCP within each program doing TCP (the program'd have to do some IPC to the IP process). This stuff started out life in PC-DOS so I don't know what grodiness is in the code because of that. >>A. I'm told that all Unix sites running 4.3 BSD also possess the MMDF >> program. PMDF was designed to talk to MMDF. > >Not true. oh, but it is true ... go look in /usr/src/new (the user-contributed stuff) and do an ls. Notice the directory which says "mmdf". Go down in there and see all 3.4 megs of source code for the system. MMDF is a nice system that's no harder to configure than sendmail and reading mmdf is a lot more enjoyable than reading sendmail configurations. -- ----- David Herron, Local E-Mail Hack, david@ms.uky.edu, david@ms.uky.csnet ----- {rutgers,uunet,cbosgd}!ukma!david, david@UKMA.BITNET ----- ----- Je parle francais comme une vache espagnole.