Path: utzoo!attcan!uunet!lll-winken!lll-tis!helios.ee.lbl.gov!pasteur!ucbvax!VENUS.YCC.YALE.EDU!LEICHTER From: LEICHTER@VENUS.YCC.YALE.EDU ("Jerry Leichter ", LEICHTER-JERRY@CS.YALE.EDU) Newsgroups: comp.os.vms Subject: Re: Press Release Message-ID: <8808022004.AA10009@ucbvax.berkeley.edu> Date: 27 Jul 88 16:14:00 GMT Sender: daemon@ucbvax.BERKELEY.EDU Organization: The Internet Lines: 94 Well, we started DEC. We bought Message Router, Digital's electronic mail transport system. Our staff is using ALL-IN-1, which uses Message Router to transport its network mail. But to make it work with DECnet mail, we had to put contorted forwarding VMSmail addresses in for everyone. But to participate in BITNET we use Jnet, by Joiner Associates, an excellent peer-to-peer DEC-IBM networking product. (Where is your product, Digital?) Jnet doesn't support Message Router. To make Jnet work with ALL-IN-1 we had to modify ALL-IN-1. We want to send mail through BITNET gateways to other networks. Jnet doesn't do it so ALL-IN-1 won't do it. Neither does VMSmail for that matter. Furthermore, we couldn't put 1200 students into ALL-IN-1 mail without buying enough VAXen to run the NYSE. So students use Dreams/6, by DCXX Software Services, an excellent personal mail system and mail transport system. It works well with DECnet and Jnet and supports RFC822, BSMTP and BITNET gateways. (Where is your product, Digital?) It doesn't support Message Router. In order to get mail from Message Router we had to modify VMSmail. Even so, a Dreams/6 user can't reply to a memo received from Message Router because the return address is munged. To participate in a TCP/IP LAN we got CMU-TEK TCP/IP, by Carnegie Mellon University. (Where is your product, Digital?) It doesn't support Message Router. Dreams/6 doesn't support CMU-TEK. What will? PMDF, from Ned Freed. (Where is your product, Digital?) It doesn't work with Message Router. At the hub of multivendor networking? Poop, I say, Digital. Poop. Let's get real here. There's a great Gordon Bell quote about standards (from well after he left Digital - it refers to the whole industry): We think standards are so great we keep creating new ones. You've listed, let's see, VMSMAIL, ALL-IN-1 MAIL/Message Router, BITNET, RFC822, BSMTP, hence IBM networking protocols, DECnet, TCP/IP. You could have listed UUCP/News, other IBM protocols (SNA, Profs), and who knows what else. No one company could possibly support all this stuff. You might as well flame at IBM for not supporting either DECnet or TCP/IP or UUCP (that would let you get rid of the BITNET stuff) or generic Unix for not supporting TCP/IP exclusively and retiring UUCP or .... You have one very legitimate complaint (the lack of compatibility between VMSMAIL and ALL-IN-1), an unfortunate legacy (VMSMAIL just kind of grew, pre- dating most current standards, and it's just too limited to play in today's world without the warts showing through) - but at least there are workarounds. TCP/IP is also a sore spot, but why do you care WHO writes the code, provided that it is available? DEC resells two 3rd-party TCP/IP packages (Wollongong and I think Fusion). There are announced standards for networks and mail systems (the ISO stuff, X.400) which DEC has announced it intends to support. In fact, it supports X.400 TODAY. Why not flame at IBM and all your other vendors for not being up-to-date and supporting X.400? As a matter of fact, DEC sells a number of products to interconnect with IBM systems. IBM systems themselves support such a multiplicity of protocols that you could come up with a similar litany of complaints just for talking to them. Much as I understand the difficulties and frustrations of building networks on top of the huge variety of "standards" in existence today, I think your complaints are mainly misdirected. If Joiner's stuff won't talk to Message Router, why don't you complain to Joiner? They are at least an equal part of the problem. If Dreams/6 won't talk to CMU/TEK, why not complain to DCXX? If PMDF won't talk to Message Router, well, I won't suggest complaining to Ned Freed - he's not paid for doing the excellent job he's done - why not get someone to write you a Message Router channel program? I'll bet it's easier to do than many of the hacks you've already got thrown together, and you'll certainly find a lot of people out there who would LOVE to have it. (They might even be willing to pay.) The ONLY way out of this mess is to chose a single set of protocols and stick with it. The current mish-mash of small fiefdoms and special-purpose hacks cannot work in the long term. There are four conceivable contenders for this kind of role: DECnet, SNA, TCP/IP, and OSI. DEC has already announced that has no intention of pushing DECnet as THE standard, but will merge with OSI. Outside of the IBM world, the chances of SNA taking over are essentially nil (though within the IBM world, it'll likely be around forever). The TCP/IP vs. OSI battle has been heated in the past, and there's no point in rehashing it; but I think even the most ardent TCP/IP supporter will have to concede that, whatever the technical merits, OSI has won (except within the academic community, and there it's probably just a matter of time). We can only hope that five, ten years from now the network interconnect problems of today will be dimly remembered by the "old fogeys" like us; the newcomers will just plug stuff in, just as we plug in appliances without worrying about AC vs. DC, 60Hz vs. 50Hz. The transition phase will, however, be painful. To expect anyone to pour large amounts of manpower and money into providing short-term hacks, when they could be trying to build stuff with a future, is foolish. -- Jerry [Who expects to be flamed by all the TCP/IP patriots and such :-)]