Path: utzoo!attcan!uunet!tut.cis.ohio-state.edu!ucbvax!INFOODS.MIT.EDU!KLENSIN From: KLENSIN@INFOODS.MIT.EDU (John C Klensin) Newsgroups: comp.protocols.ibm Subject: Re: re: e-mail and System/370 Systems Person Message-ID: <9004111823.AA18618@lilac.berkeley.edu> Date: 11 Apr 90 14:08:27 GMT Sender: daemon@ucbvax.BERKELEY.EDU Reply-To: John C Klensin Organization: The Internet Lines: 37 Jeff, I'm going to post this to the list on the theory that the principle, if not the specific details, may be of general interest (again). >I'm told that a better form of my e-mail address is as follows: > jbeard%quintus.com@sun.com "quintus.com" is a registered Domain Name System name on the Internet. This makes the "better" form of your e-mail address (as well as the simple and obvious one) jbeard@quintus.com Now, where does "sun.com" come in? At the moment, mail addressed to quintus.com (as above) from the Internet is sent--automatically, by the mail sending software--to "sun.com" which acts as a Mail eXchanger for "quintus.com". But, users sending mail to you don't need to know that, and *shouldn't* know and remember that. Why? Well, if quintus ever changes its connectivity, or an additional Mail eXchanger becomes available, this can be reflected in easily updated (distributed and dynamic) tables, retaining the name, while use of %quintus.com@sun.com specifies a connectivity and a routing, which is, in general, a bad idea. This address should also work from BITNET, as the use of the domain ".COM", if not found otherwise, will route the mail to a BITNET->Internet gateway. All of the correctly-functioning ones of those are able to figure out what to do with "@quintus.com" and do it; any that are not correctly-functioning should be corrected. Generalization and moral: Use addresses of the form user%host@domain only when absolutely necessary, i.e., when "host" is *not* a registered Internet domain name. If "host" is a registered Internet domain name, use of the user%hostdomain@gatewaydomain form, rather than the user@hostdomain form, does not add any addressing or performance advantages, and tends to cause long-term addressing problems as networks and connections evolve. --john Klensin@MIT.EDU -------