home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!cis.ohio-state.edu!pacific.mps.ohio-state.edu!linac!att!ucbvax!ALW.NIH.GOV!rdew
- From: rdew@ALW.NIH.GOV (Bob Dew)
- Newsgroups: comp.soft-sys.andrew
- Subject: Re: "slow networks"
- Message-ID: <MebPm_m0ts4j4mfEtI@alw.nih.gov>
- Date: 28 Aug 92 05:41:30 GMT
- References: <IebJxxoGf0478PRKEF@athena.mit.edu>
- Sender: daemon@ucbvax.BERKELEY.EDU
- Distribution: world
- Organization: The Internet
- Lines: 19
-
-
- AFS may not be for everybody. We use it for reasons of security,
- scalability, and ease of administration. Likewise, we find AMDS secure,
- scalable, and easy to administer and use under AFS. Unfortunately,
- messages, cui and vui are the only user agents capable of handling AFS
- home-area mail delivery. To remove the message server code from ATK
- would mean, for us, the end of distributed mail service under AFS, and a
- return to a client-server model that we think is inherently prone to
- central points of failure, that is susceptible to unauthorized access
- and to catastrophic data loss, and that doesn't scale well.
-
- But regardless of what mail transfer agent a site elects to use (clearly
- there are advantages and disadvantages to both), what merits would be
- gained in the messages and vui programs by dropping ATK's message server
- support? As things are, if ATK is compiled without the options for AFS
- and AMDS, are the resulting programs much larger and more complicated
- than if the message server code didn't exist at all?
-
- -Bob
-