Robert.Ullmann@en-c06.x400.prime.COM (07/19/89)
Steve, I would like to apologise for some of my comments. I hope we can return the discussion to technical issues. And (unfortunately) political ones. I did in fact file your note appropriately; it is in my X.400-gate/recvd-from/private/Kille folder. I seem to have given some people the impression that I had discarded it ... -------- Yes, I am "energetic"! I am very exited by the prospect that we can really have a world-wide mail system, with fairly simple domain addressing. The technology has been available (in various forms) for over a decade; what has been lacking is a perception (by management) that it is needed. My experience in implementing mail in several organizations and sub-organizations, including one large multi-national corporation, is that mail is viewed as useless until after it is operational. Then the same people can't live without it. The idea that there is real world-wide recognition that *connected* electronic mail is needed by modern organizations has been a long time coming; and I am quite excited that it is finally arriving. -------- Recent message(s) to this list have lamented that the Internet has not defined how it will set up its ADMD(s) and/or PRMD(s). I believe the answer is that the Internet community is not interested in a new structuring: we already have a domain system that works. All we really need is a single-point connection in the hierarchy. This is why I am suggesting the /C=XA/ method; it is the simplest possible connection. (Even simpler would be /C=COM/, etc, with a *totally* uniform translation; this, however only works with some X.400 s/w. I did prototype it just for fun ... :-) To (hopefully accurately) quote Albert Einstein: "A theory should be as simple as possible, and no simpler." In return, the Internet will hide all of its internal structure (other than in the ISO-3166 domains), allowing the X.400 administrations to use the simple rule /C=XA/ goes to (nearest) internet gate and it will be handled. Or to the nearest X.400 MTA within the internet set up as /C=XA/... which can obviously find a gate. (such as EN-C06.Prime.COM, which is also configured to be /C=XA/A=COM/P=Prime/O=X400/OU=EN-C06/) The internet will then also route for /C=XA/A=BITNET/ and /C=XA/A=uucp/ (which gives both of those nets a standard place in the domains until all of their systems have "real" domain names. Whatever "real" means. <grin>). This also provides a place for any other foreign net, if it is properly gated to the internet as a top-level domain, whether in DNS (e.g. INT) or not as long as it is commonly routed for (e.g. UUCP). Of course "COM" isn't an "ADMD", but that doesn't matter: the internal structure of XA is not visible from outside. (Isn't that the whole point of domains?) All of which does not preclude local re-write rules. To use an example, which was kindly provided by Piet: The internet host cwi.nl is, in X.400, within /P=surf/A=400net. When this is translated with the Dutch gateway rule, either in nl, or anywhere else that has set up the rule: /C=nl/A=400net/P=surf/O=cwi/ cwi.nl when translated elsewhere, this results in: cwi.surf.400net.nl which will follow the *.nl or *.400net.nl MX route, and can then be translated (even if only by looping it through the Dutch gate!) The reverse case will result in /C=nl/A=cwi/, which is handled by the X.400 MTA by simply throwing any unknown C/ADMD pair at a gate, and letting the other side look it up in DNS. (which, at least until X.400 is self-connected as well as the internet, is what the MTAs will be doing anyway.) This will be particularily easy for MTAs such as the one Christian Huitema referred to which access both systems with equal ease. This means that knowledge of others' translation rules becomes a convenience, not a necessity. Which makes the administration of the Pan-European Master Table (:-) much easier; it can be updated and distributed as desired by the participants. Proposed entries for the Internet: $ADMD=<name>.$C=XA#<name># and <name>#$ADMD=<name>.$C=XA for <name> in { COM, EDU, NET, MIL, GOV, INT, IRL, ORG, UUCP, BITNET, ARPA } (my gate has this built-in, for any "top-level" name > 2 letters, but that is just a refinement that anticipates any domains that might be added later) And then maybe we can talk about technical details, like getting more local-parts and personal names to reversibly map? Robert
mark@cblpf.att.COM (Mark R Horton) (07/20/89)
While I still don't think this proposal is realistic, may I point out one possible improvement anyway: Change /C=XA/ to /C=US/. The rest of the world has long told us that they view the big six (COM, EDU, GOV, MIL, ORG, NET) domains as if they had a .US on the end of them, anyway. With the exception of a limited number of Canadian sites (which I think are moving to .CA) the only people to register in these domains have been in the US anyway. If you use /C=US/ADMD=COM/PRMD=org/etc notion, then you only have to get the US X.400 carriers to buy into the notion of 6 extra USA ADMDs to be passed to the gateway. The current proposal requires EVERY WORLDWIDE X.400 carrier to support your XA notion and route it to an Internet gateway. Personally, I don't believe either buy-in option is very likely. The X.400 people I've talked to have the attitude "TCP/IP? We don't care about them. X.400 is an international standard, we support it.", "We provide service to our customers for internal email, and we're working on a few interexchange agreements with some other vendors.", and "If we get incoming email from the Internet, whom do we bill for it? We won't pass mail we can't bill for." Mark
Stef@NRTC.NORTHROP.COM (Einar Stefferud) (07/20/89)
Hi Mark, et al ... Your points are well taken. A number of us have been thinking about all this for a long time now, and we have come all the way around to realizing that the big six (COM, EDU, GOV, NET, ORG, MIL) are really operating at the PRMD level. In fact, the entier Internet Mail Servie is operating as a PRMD in that it is not offering services to all who offer to pay. It is a Private network and it is a Management Domain. So, we are looking toward ways to organize the Internet Top Level Domains for X.400 purposes. This is not a imple task, in that each of the Big Six must be considered separately. WE cna imagine the EDU community organizing and choosing a name that corresponds to EDU for their PRMD name. We do not see the Commercial Comunity choosing to operate PRMD=COM, since many of them will wnat their own PRMD anyway. We expect that PRIME will want to do this, along with DEC, IBM, XEROX, et al... In any case, the internet wizards cannot choose for them! WE can see a home for GOV and MIL. NET is small and we expect that they can figure out what to do for their X.400 PRMD name. But, the fact is that the decisions cannot be imposed by the internet for purposes ofd simplifying the mappings because the Internet does not have the required authority to impose any such decision. Yes, it woudl be nice if we could simply dictate this sort of thing, but then we would probably not like the choices made by the appointed dictator. I think it is safe to expect that we would not be named to this exhalted position. So, I and others have become convinced that the next step must be for the various Internet Communities to decide how their registered members wish to register themselves in the X.400 namespace, if and when an X.400 namespace authority materializes. My best reading on this score at this time is that Judge Greene awarded responsibility and authority for all naumering and anming plans in the US to BELCORE. I have made an enquiry to BELCORE but do not yet have any response. However, if we can get everyone to agree on selecting a dictator, you re welcome to select one. I will elect to pass however, since I do not subscribe to the "1984" view of how to do these things. Recent events in China help to reinforce my views. Cheers...\Stef
Christian.Huitema@MIRSA.INRIA.FR (Christian Huitema) (07/20/89)
There is certainly one inconveniency in all those routing tables: if they are not small and stable, e.g. with half a dozen entries per countries, then they wont be up to date, and we must guarantee either an ubiquitous way of accessing them or a default solution. The ubiquitous way is indeed the DNS, through the creation of a new type of field (X400 ?): when queried with a domain name, it would respond with the equivalent X.400 domain. The default solution can also be based on the domain name. Print the orname component as a domain name (OU.O.P.A.C), apply the well known stable mappings in order to reduce to some (OU.O.Top) or (OU.O.P.Top) and then look for the appropiate gateway in the MX records. Or for a CNAME record. It may work. Resulting requirement: if a <domain> acquires an X.400 subscription, it has to enter a new <X400> record in the DNS, and an MX record for the X400 domain. For example, if IBM.COM gets a subscription from MCI-MAIL, it will have to provide: MX: IBM.COM => IBM gateway; IBM.MCI-MAIL.US => the same IBM gateway; X400: IBM.COM => IBM.MCI-MAIL.US There is no much cost for the existing domains, except indeed for the dreadful ".US" domain which will have to cope with the recording of ADMDs... Christian Huitema
mark@cblpf.att.COM (Mark R Horton) (07/21/89)
Stef - you say the Internet is a PRMD because they won't offer service to all comers. Yet, we seem to find that the regional supercomputer centers, which apparently also offer regional TCP/IP Internets with Internet connectivitiy, pretty much do offer service to all comers. Sounds to me like the Internet is an ADMD, not a PRMD. Am I missing something? Mark
Stef@NRTC.NORTHROP.COM (Einar Stefferud) (07/21/89)
Hello Christian -- I don't see how .US with its STATE oriented substructure is any more "dreadful" than .DE with its missing ORG units, or .FR with ... On the other hand, I believe it has always been one of the objectives of RFC987 to support the use of the DNS records (e.g., X400 and MX) in ways such as you describe. I expect that this is fully preserved in the latest version of RFC987(88). Best...\Stef
Stef@NRTC.NORTHROP.COM (Einar Stefferud) (07/21/89)
The NSFnet Regionals offer membersips to all, but they offer services only to their members. Thus, we don't see the Internet competing with the commercial carrier directly, but offering PRMD services to its "members". I realize that these distintions tend to be somewhat tenuous and fragile, but none-the-less, if I had to decide which role the Internet plays, with no direct charging scheme for traffic among its members, and comparing that with ATTMAIL, MCIMAIL, TELEMAIL, DIALCOM, et al, I conclude that the Internet is not an ADMD. The next question arises on whether the ADMD's per se will deal with the INTERNET as an equal. I expect not. The ITNERNET has not participated in the IAOG negotiaons for inter-carrier settlements, and is not involved in any way in the Aerospace Industray Association negotiations with the US ADMDs regarding ADMD interconnection, etc, et al. I just don't see any way for the Internet to assume a role as an ADMD. For example, who would speak for it in the above mentioned negotiations? No, it actually simplifies lots of issues if we just relax and accept tha the Internet is actually operating as a PRMD, with some negotiated cnnectins with various ADMDs, including UUNET. ATTMAIL has connections, but has not negotiated any kind of an understanding with the Internet. Actually they have been seriously trrying to break all the links for the last 3 years. Very nearly succeded. At least they have rendered the interconnect sufficiently useless as top cause me to remove all ATAMIL addresses from any of my mailing lists. Too much failed mail! The way things look in the US, the big six (COM, GOV, EDU, MIL, ...) will probably each go theeir own way as logical PRMDs, though inside the Internet they will continue to be mostly fully connected IP sites, with some RELAY connections like UUNET, CSNET and BITNET, et al. I see a logical PRMD being a agency that administers the name space and routing tables to rpovide information to others for how to reach it, and its "members". Best...\Stef
piet@cwi.nl (Piet Beertema) (07/21/89)
The X.400 people I've talked to have the attitude "TCP/IP? We don't care about them. X.400 is an international standard, we support it.", "We provide service to our customers for internal email, and we're working on a few interexchange agreements with some other vendors." That's indeed the ivory tower point of view that is so characteristic of a lot of X.400 (and OSI, for that matter) people. It's the point of view that is destructive to good internetworking. It's also the point of view that completely disregards the NEED of a very large part (mostly academic) of the X.400 "community" to talk to the Internet and other networks. That's why a lot of their socalled "internal email" will in fact be external email for a long time to come! Piet
CERF@A.ISI.EDU (07/22/89)
The Internet can be and will be linked to the public email carriers. The terms and conditions for such linkages are being worked out, largely with the members of the FRICC. For the most part, the elements of the Internet serve a research and development-oriented community. The general access and use policies for those parts of the Internet which are operated under government subsidy tend to restrict use to that which is supportive of the R&D objectives or operational objectives of the government sponsor. In terms of connectivity, the Internet probably should not be considered an ADMD if this implies that it will act as a transit facility for other public ADMDs. An element of Internet use policy tends to be exclusion of commercial transit traffic (e.g. Telemail and Dialcom should not send traffic through Internet to each other - the relays linking public mail to the Internet should block such traffic. If I have correctly surmised that ADMDs are expected to be able to negotiate a transit use/charge policy, then the Internet probably should be treated as a PRMD. In the long term, however, it is my understanding that the FRICC would like to see many of the services now available on the Internet also available commercially so that Internet community users could have the option of obtaining service from commercial carriers. Comments on this view would be most welcome. Vint Cerf
Stef@NRTC.NORTHROP.COM (Einar Stefferud) (07/22/89)
Before someone suggests that this is not the right list for disucssion of Vint's question, I want to endorse it for discussion. The fact that this is an international list should not inhibit this discussion. I expect that other countries have similar issues to resolve, and we in the US would appreciate any experience or suggestions from abroad. Best...\Stef
lubich@ethz.UUCP (Hannes Lubich) (07/25/89)
[Piet writes:] >That's indeed the ivory tower point of view that is so >characteristic of a lot of X.400 (and OSI, for that matter) >people. Bullshit. X.400 connectivity in most organizations is based on DECNET and TCP/IP while, due to costs and unavailability of an Internet in Europe, wide area connectivity between X.400 domains is based on X.25 and/or leased lines. Since nobody wants to support different X.400 products just to cover both LAN and WAN connectivity, we *DO* care about TCP/IP. We don't see TCP/IP as the one and only solution, however. >That's why a lot of their socalled "internal email" >will in fact be external email for a long time to come! Nobody used this term (except from you, perhaps). X.400 systems have been connected to the Internet from quite some time now (I've been running a gateway from the Swisswide academic X.400 network to CSNET since 1986, for instance). RARE MHS connects you to about 50 different networks, *including* all Internet domains, so what's the point here? Piet, why do you continue to willingly post misinformation, despite the fact that you should know better and that this won't help anybody? Could we go back to the technical points of this discussion, please. Regards --HaL -- ~ UUCP/Usenet : {known world}!mcvax!cernvax!ethz!lubich ~ or : lubich@ethz.uucp ~ CSNET/ARPA/BITNET : lubich@inf.ethz.ch / lubich%inf.ethz.ch@relay.cs.net ~ The usual disclaimer : No, it wasn't me, somebody must have used my account.
piet@cwi.nl (Piet Beertema) (08/01/89)
X.400 connectivity in most organizations is based on DECNET and TCP/IP while, due to costs and unavailability of an Internet in Europe The latter is rapidly changing. X.400 systems have been connected to the Internet from quite some time now (I've been running a gateway from the Swisswide academic X.400 network to CSNET since 1986, for instance). Yes, and right here at the heart of EUnet I've seen all the problems and failures caused by numerous ad hoc (named "official" though) gateways passing by. Some time ago it was stated that "European e-mail is a mess"; well, it wasn't before all those X.400 gateways came into existence! RARE MHS connects you to about 50 different networks, *including* all Internet domains, so what's the point here? Network congestion, address mangling, unwillingness to cooperate with other networks, etc. etc. That's the point here. Piet, why do you continue to willingly post misinformation, despite the fact that you should know better and that this won't help anybody? I'm not posting any misinformation whatsoever, I'm just presenting facts, facts that some communities (e.g. RARE) prefer to close their eyes for. Ivory tower is a correct description of such an attitude. Piet