Newsgroups: comp.archives.admin
Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!src.honeywell.com!msi.umn.edu!s16.msi.umn.edu!zoo
From: zoo@aps1.spa.umn.edu (david d [zoo] zuhn)
Subject: Re: bitftp changes, alternatives
In-Reply-To: ray@vantage.UUCP's message of 13 Jun 91 22: 53:57 GMT
Message-ID: <ZOO.91Jun16121555@phacelia.spa.umn.edu>
X-Md4-Signature: bafb8ed15603312fd7a44242ef3a2689
Sender: news@s1.msi.umn.edu
Nntp-Posting-Host: aps1.spa.umn.edu
Organization: Minnesota Automated Plate Scanner Lab
References: <121080001@vantage.UUCP>
Date: Sun, 16 Jun 1991 18:15:55 GMT
Lines: 41

>>>>> On 13 Jun 91 22:53:57 GMT, ray@vantage.UUCP (Ray Liere) said:
Ray> I would be interested in someone/s in the know summarizing what led
Ray> to the change -- I assume that too much success did them in?

The big problem with bitftp and other mail based servers is that they
tend to go to low-bandwidth mediums of communications, with
intermediate links between the source and the destination.  If your
site talks directly to an internet node from whence your mail comes
from, bitftp doesn't use anyone's resources other than your own.

Unfortunately, this often isn't the case.  If there is any site
between your machine and the internet link, that site gets to pay for
your mail.  This is usually just fine when mail consists of a few
kbytes per day, but when GNU emacs or GIF files or many other large
files are requested, the links become jammed.

For example, CompuServer connects to the Internet via a phone link
(9600 or 19200 baud) to Ohio State.  One day, notice of the bitftp
server was posted to CI$, and the next day over 50Mb of files was
enqueued at OSU for CI$, mostly GIF files.

This use of other people's equipment is the reason that bitftp and
other mail-based servers have been shut down or had their service
restricted to their original purpose (BITnet/Internet in the case of
bitftp).

Provision of future mail based servers is unlikely, given the current
feelings of several major archivists.  The correct solution is for
your site to connect to one of the anonymous UUCP archives and
download what you want yourself.  This places the onus of the
responsibility on the people who want the software, not the
intermediate links.

UUCP access is provided by OSU and UUnet, both for the cost of the
phone call.  I don't have addresses handy, but the OSU notice is
posted often to comp.sources.wanted.  

(I'm not trying to denigrate you personally at all, I'm just using you
as an example).  


