Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!samsung!uunet!bfmny0!tneff From: tneff@bfmny0.BFM.COM (Tom Neff) Newsgroups: comp.mail.misc Subject: Re: mail servers Message-ID: <15781@bfmny0.BFM.COM> Date: 22 Aug 90 08:02:39 GMT References: <1990Aug17.021921.1863@chinet.chi.il.us> <1594@i2ack.sublink.org> Reply-To: tneff@bfmny0.BFM.COM (Tom Neff) Lines: 41 In article bob@MorningStar.Com (Bob Sutterfield) writes: >A mail-based archive server is an impolite and destructive way to >distribute large amounts of information, and should be discouraged in >such applications. Almost -- a MBAS is *potentially* destructive if it imposes no bandwidth restrictions, satisfies requests exclusively by mailing files back to the requestor, and makes no allowance for the loading constraints on the delivery path. This gloomy potential need not be realized. * Responsible archive servers will establish and enforce limits on how much bandwidth an individual user can have; how much a given reply path can tolerate; and how much overall activity per day can be supported. * In the most general sense of the term, a mail based archive server is any server that accepts requests via mail. The actions taken to satisfy those requests need not be limited to mailing file contents back along the reply path. - A server might, for instance, satisfy a mailed file request by placing the requested file on the nearest Internet site, and mailing a short notification of FTP availability back to the requesting user. - It might respond to a mailed request by placing the file in an anonymous UUCP download directory, and mailing back L.sys and filename information to the user. - It might log the request, collate it with others for the same file, and eventually perform one single contents mailing to a regional site, followed by a mass mailing to the requesting users in that region notifying them of local availability. * Where mailing the file is unavoidable, the server might identify the message content as such for the management convenience of intervening sites. Exiting headers like Precedence: and Content-Type: could be used. Since the MBAS genie is not going to go quietly back into its bottle, it would be more constructive to suggest responsible strategies for running them -- and provide software to implement those strategies -- than to inveigh against the entire phenomenon. -- The real problem with SDI is %/ Tom Neff that it doesn't kill anybody. /% tneff@bfmny0.BFM.COM