reid@decwrl.dec.com (Brian Reid) (08/03/88)
This evening, after weeks of behaving itself, decwrl's pathalias database told me that the best route to mcvax from here is sun!daver!altnet!altmail!altger!unido!mcvax!%s This is completely preposterous. If my MTA started routing mail along paths like this I would newfs it and start using the telephone. I'm sure that if I take the trouble to go look (I probably won't) I'll find that according to the current pathalias data this really *is* the best path from decwrl to mcvax. It's just that pathalias and I have different ideas of what constitutes "best". Remember, folks, pathalias believes exactly what you tell it. Until problems like this stop happening, MTA's should not do routing.
lmb@vsi1.UUCP (Larry Blair) (08/04/88)
In article <676@bacchus.DEC.COM> reid@decwrl.UUCP (Brian Reid) writes:
=
=This evening, after weeks of behaving itself, decwrl's pathalias database
=told me that the best route to mcvax from here is
=
=sun!daver!altnet!altmail!altger!unido!mcvax!%s
Since we talk to sun, daver, and altnet, let's run "address xxx@mcvax":
xxx@mcvax: altnet!uunet!mcvax!xxx
This is why I always hand run pathalias. A few days usually don't matter
and I can be sure that my database is complete.
Another thought occurs to me: Do you have "-d uunet" in your pathalias
command?
=This is completely preposterous. If my MTA started routing mail along paths
=like this I would newfs it and start using the telephone. I'm sure that if I
=take the trouble to go look (I probably won't) I'll find that according to
=the current pathalias data this really *is* the best path from decwrl to
=mcvax. It's just that pathalias and I have different ideas of what
=constitutes "best". Remember, folks, pathalias believes exactly what you tell
=it.
=
=Until problems like this stop happening, MTA's should not do routing.
There is nothing wrong with letting the MTA route for you if the first hop
is a non-connect. Your mail WILL get to mcvax, unless (pet peeve!) it goes
through some Bozo site, like rutgers, that thinks that the maps are gospel
and refuses to let you do your own routing. As long as the MTA allows
specific routing, it is far better to concentrate route generation in the
one facility.
Btw, our bounce rate for "reasonable" paths has gone to zero since I added
"-d rutgers" to our pathalias run.
--
* * O Larry Blair ames-----\
* * O VICOM Systems Inc. pyramid---\
* * O 2520 Junction Ave. uunet!ubvax!vsi1!lmb
* * O San Jose, CA 95134 sun-------/
* * O +1-408-432-8660 +1-800-538-3905kehres@tis.llnl.gov (Tim Kehres) (08/04/88)
In article <676@bacchus.DEC.COM> reid@decwrl.UUCP (Brian Reid) writes: > > This evening, after weeks of behaving itself, decwrl's pathalias database > told me that the best route to mcvax from here is > > sun!daver!altnet!altmail!altger!unido!mcvax!%s > > This is completely preposterous. If my MTA started routing mail along paths > like this I would newfs it and start using the telephone. I'm sure that if I > take the trouble to go look (I probably won't) I'll find that according to > the current pathalias data this really *is* the best path from decwrl to > mcvax. It's just that pathalias and I have different ideas of what > constitutes "best". Remember, folks, pathalias believes exactly what you tell > it. > > Until problems like this stop happening, MTA's should not do routing. Yes, this is a preposterous route, especially considering you are an internet site. It is also highly likely that pathalias thinks that this is a "best" route from your site to mcvax. It is also quite true that you can get to mcvax in just a few hops from where you are. If you were to look into what pathalias generates, I am sure that you could come up with countless other examples of how the routing could be done in a more efficient and cost effective way. This is not however a valid reason for not letting MTA'a handle routing, whenever an explicit route is not specified. Some of the reasons most modern mail systems are evolving to domain based addressing are: o Optimal source routing can be dynamic and the originator/recipient cannot always determine best routing. o Most users are only concerned about getting their mail through, not how it gets there. When is the last time you specified the route the postal service should use to get a letter delivered? How many of your users can tell you the best route to mcvax without looking through past mail messages or uucp tables? o The use of a common set of domains makes the use and implementation of gateways between different electronic mail systems easier. This is one of the reasons for the establishment of standard domain names. These are probably not the only reasons, but enough to make the point. There is sometimes a tradeoff between best routing and making the use of electronic mail possible for a wider range of users. Although pathalias (or the data that gets fed into pathalias) certainly can be improved, it does help (most of the time) in generating good paths, and provides some isolation to a wide range of users to routing decisions. Tim Kehres Control Data Corporation / Lawrence Livermore National Laboratory MILNET: kehres@tis.llnl.gov AT&T: (415) 463-6852
chip@ateng.UUCP (Chip Salzenberg) (08/04/88)
According to lmb@vsi1.UUCP (Larry Blair): >Btw, our bounce rate for "reasonable" paths has gone to zero since I added >"-d rutgers" to our pathalias run. Amen! Both rutgers and harvard are marked as "dead" in my local adjustments file, which is automatically included by my pathalias wrapper script. I personally think that rutgers should be dropped from the net unless the arrogant "you can't possibly know more about routing than me" attitude is replaced by something more cooperative. Unfortunately, too many sites depend on rutgers for this to happen. If you agree, send mail to Mel Pleasant <pleasant@rutgers.EDU> and complain. If Mel gets enough flames, he just might reconsider. -- Chip Salzenberg <chip@ateng.uu.net> or <uunet!ateng!chip> A T Engineering My employer may or may not agree with me. You make me wanna break the laws of time and space You make me wanna eat pork
brisco@pilot.njin.net (Thomas Paul Brisco) (08/05/88)
In article <881@vsi1.UUCP> lmb@vsi1.UUCP (Larry Blair) writes: > > There is nothing wrong with letting the MTA route for you if the first hop > is a non-connect. Your mail WILL get to mcvax, unless (pet peeve!) it goes > through some Bozo site, like rutgers, that thinks that the maps are gospel > and refuses to let you do your own routing. As long as the MTA allows > specific routing, it is far better to concentrate route generation in the > one facility. > > Btw, our bounce rate for "reasonable" paths has gone to zero since I added > "-d rutgers" to our pathalias run. > -- > * * O Larry Blair In fact, there is nothing wrong will letting your MTA do all the routing for you. I've had very few sites bouncing back to me, and I _usually_ route through rutgers. I do get a few bounced now and then -- but I attribute that to brain-dead administrators who can't be bothered to keep their maps up to date. You can hardly blame rutgers for letting pathalias do *all* of the routing -- the maps are supposed to mirror the actual connectivity available (but are also used to add "administrative" weights at some sites). I've had one site directly off of rutgers for about a year now (a machine at home), and currently have a second one with a lot of other sites in addition. I really don't want to spend my time figuring out what is the quickest path between sites X and Y, I really think my attention is best put in other places. What I *do* have a problem with is the turn-around time for a map to get published, but since I really can't think of a better method for distributing the maps I'm not going to complain too loudly - and the turn-around time isnt *that* bad. What I *really* get ticked with is the administrators who can't be bothered with keeping their maps up to date. Simply because Rutgers chooses to follow the maps _as published_ is hardly a reason to flame on them. *THAT IS WHAT THE MAPS ARE THERE FOR*. I've seen sites that were dead for about a year without the silly-a**ed administrator bothering to remove the map, *nor remove the entry from his/her other hosts maps*. Even after flaming at them periodically for months, they still could not find time to delete a couple of lines from their existing map entries. You would do better to spend time flaming the sites that cannot spend the 15 to 30 seconds that it takes to update the maps as necessary - NOT to flame the sites that implement the maps as intended. Tp. "In this perfect world ..." -- ---------------------------------------- UUCP: ...!rutgers!brisco ...!njin!brisco ARPA: brisco@pilot.njin.net BITNET: brisco@zodiac.bitnet Voice: (201) 932-2351 ---------------------------------------- Ask me about the New Jersey Intercampus Network Pilot Project
bill@ssbn.WLK.COM (Bill Kennedy) (08/05/88)
In article <340@ateng.UUCP> chip@ateng.UUCP (Chip Salzenberg) writes: >According to lmb@vsi1.UUCP (Larry Blair): >>Btw, our bounce rate for "reasonable" paths has gone to zero since I added >>"-d rutgers" to our pathalias run. > >Amen! Both rutgers and harvard are marked as "dead" in my local adjustments >file, which is automatically included by my pathalias wrapper script. I don't often disagree with Chip but he's solved the problem with his local adjustments, as he reports. >I personally think that rutgers should be dropped from the net unless the >arrogant "you can't possibly know more about routing than me" attitude >is replaced by something more cooperative. Unfortunately, too many sites >depend on rutgers for this to happen. Here I whole heartedly disagree. Any site is, of course, free to route around rutgers and I used to contribute a few db/btu's to this topic, but "removing" rutgers isn't the answer at all. 1) By _definition_ rutgers is the official map custodian 2) Any site is free to fold, spindle, or mutilate mail or news 3) Rutgers is trying to improve the situation and improvement attempts are easy to mistake for malice. The best way to encourage sites to keep their maps up to date is to have other sites use them. That's precisely what rutgers does. When I see mail jostled all over the country because of a screw up in ssbn's map, a new map is sent in promptly. Such wierdness is normally a result of rutgers' literal interpretation of ssbn's map, ssbn is mistaken, rutgers is not. If you want to suggest that the maps are wrong, you are mistaken. The map that's posted is, by definition, the way that site wants to look to the rest of the net. That's why many sites have their posted map and another that they use for themselves to generate their own paths. >If you agree, send mail to Mel Pleasant <pleasant@rutgers.EDU> and complain. >If Mel gets enough flames, he just might reconsider. >-- >Chip Salzenberg <chip@ateng.uu.net> or <uunet!ateng!chip> >A T Engineering My employer may or may not agree with me. > You make me wanna break the laws of time and space > You make me wanna eat pork That's doubtful too Chip, he's got some pretty flame retardant garb and I'm pretty sure that flames on this topic will get to /dev/null more directly than they got to rutgers :-) If their behaviour is objectionable you can do as you have done, route around them. I think that the ultimate solution is for the posted maps to reflect how each site really wants mail to move. If that was the case we would not be having this discussion. Let me briefly describe a scenario in support of this. There are ssbn neighbors who log in from time to time, directly, to collect some bulky item that shouldn't be mailed. They appear in ssbn's map in order to report that such a connection exists, not to advertise it as a good path. It has a very high cost on it to discourage pathalias but it exists in case a router can't find any better path to get it there. In those few cases ssbn will gladly forward the mail but it is not the "preferred" route. If mail for that site came from rutgers (ssbn is a rutgers neighbor) I am *certain* that there was no better way, so off it goes. When we see some beax eaux routing done by rutgers you can be sure that it's the result of some beaux eaux map, not malicious intent by rutgers. Roasting Mel won't cut it, we have to use peer pressure to make people get it right before they send it in. Gosh! Look at the agony AT&T is in with their new (monster?) approach. They can't get their own SA's to get in step and it's causing near cardiac size problems within att. Finally (mercifully), when I don't know what else to do with a piece of mail I toss it at rutgers. If it bounces from them I can be sure that it just can't be delivered within the net as posted. I think that's a good use of rutgers even if you don't like what they do as a matter of routine. -- Bill Kennedy usenet {killer,att,rutgers,sun!daver,uunet!bigtex}!ssbn!bill internet bill@ssbn.WLK.COM
aad@stpstn.UUCP (Anthony A. Datri) (08/06/88)
>Amen! Both rutgers and harvard are marked as "dead" in my local adjustments >file, which is automatically included by my pathalias wrapper script. I can echo the sentiment on harvard. Much of what I sent anywhere near them gets thrown to husc6 for some unknown reason, and as far as I can tell husc6's mailer doesn't even look at the address, it just bounces *everything*. Not to mention the incoming mail that I've seen with ...!harvard!yale!harvard in it's path -- @disclaimer(Any concepts or opinions above are entirely mine, not those of my employer, my GIGI, or my 11/34) beak is beak is not Anthony A. Datri,SysAdmin,StepstoneCorporation,stpstn!aad