robert@uop.edu (Fred Flintstone) (01/20/88)
I recently had the "excitement" of being part of a mailer grump that was the result of the use of a path in our uucp/paths file. I was testing possible paths to a friend in Southern California, and had used two workable, but average paths. So, I thought I would see what path was in our entry for that site. Low and behold, the mail path was fewer machines, so I tried it. Unfortunately, it blows through uunet from lll-winken, and guess what? Yep, I got into some trouble due to fees for the link from uunet to the So.Cal. machine that hooks up to my friend. Normally, this might have occurred to me, but this time it did not. My question is this: Are the files set up in a path alias group arranged for speed? (this would make sense in one line of thinking). Why not provide an alternate, like flying "coach". Had I mailed to user@system, it would have defaulted to the uunet path anyway, and that seems abit unclean in the light of charges, and in the light of the various west coast systems I can use to gain access. I plead ignorance for using the "suggested" path, but does a better way exist? If I were a user that did not know the number of sites and connections, then surely the problem would have been compounded by mailing to user@system. Is there a trade off for time and money? Surely, an automated map update like that is costing someone something!! My apologies to this group if it is not the right place for such a question. Please respond by e-mail where economical as I don't always get all of this group. Thank you.
wendyt@pyrps5 (Wendy Thrash) (01/21/88)
In article <923@uop.edu> robert@uop.edu (Fred Flintstone) writes: >So, I thought I would see what path was in our entry for that site. >Unfortunately, it blows through uunet from lll-winken, and guess what? >Yep, I got into some trouble due to fees for the link from uunet to >the So.Cal. machine that hooks up to my friend. I've always believed that the costs we put in our map entries should reflect not only the frequency and speed of the link, but also our willingness to forward mail over it. Any site that advertises a DIRECT link to another site shouldn't be surprised when mail starts flowing over it. At unisoft I routinely adjusted map entries so we wouldn't suddenly become California's gateway to Japan or England. I used a slightly different version of our map entry locally, so we wouldn't send mail for our distributors through some roundabout path. The use of two different versions of one's map file is inelegant, but effective. --- Wendy Thrash, now ...!pyramid!wendyt
woods@hao.ucar.edu (Greg Woods) (01/22/88)
In article <13506@pyramid.pyramid.com> wendyt@pyrps5.UUCP (Wendy Thrash) writes: >I've always believed that the costs we put in our map entries >should reflect not only the frequency and speed of the link, but >also our willingness to forward mail over it. I really would have thought that this was self-evident, but the fact that we are actually having this discussion proves me wrong. The pathalias cost is inversely proportional to the likelihood of other sites using that link if you publish it. And if you really don't want anyone but locals using a link, then don't put it in your map entry at all. >The use of two different >versions of one's map file is inelegant, but effective. You actually don't need two versions of the entire map. Only the links that you want different locally then in your official map entry (we have several like that) need to be in your second copy. If pathalias finds two different costs for the same link, it assumes the cheaper one. Therefore you can have an official map file and a set of local changes (as long as those changes all reflect new links or cheaper costs for existing links) both in the pathalias input and it will work out fine. --Greg