[309] in Commercialization & Privatization of the Internet
A few questions re current discussions...
daemon@ATHENA.MIT.EDU (Stephen Wolff)
Wed Mar 6 13:57:27 1991
To: ddern@world.std.com (Daniel P Dern)
Cc: com-priv@psi.com
Date: Wed, 06 Mar 91 13:25:39 EST
From: Stephen Wolff <steve@cise.nsf.gov>
> 2. Is anonymous FTP NSFnet abuse?
>
> Only where the NSFnet backbone is used inappropriately, no? It's
> unreasonable to expect people to remove this service, or not use it,
> within their own nets, regionals, commercials, uucp-based and other
> servers, etc. Hey, maybe policy based routing would help! Or more
> of them unrestricted networks! Or we could install parking meter or
> other coin collection boxes on the routers :-)
More unrestricted nets is the correct answer. Until that happens there is
another option: a sysop in bull.com has very kindly made available a hack of
ftp and traceroute that can be installed by the administrator of a host
admitting anonymous ftp. In response to ftp requests for any set of files,
the s/w does a traceroute on the requestor and if the route contains any
member of an adminstrator-selectable set of "forbidden" networks (e.g., the
NSFNET Backbone) it politely declines to make the transfer. Not perfect,
but easy. C source available on request from NSF, without warranty, etc.
> 3. Is there life beyond FTP, Telnet, SMTP
>
> What do folks have in mind, in terms of new services... I'm not arguing,
> I'm asking....
Well, yes. You probably haven't seen the half-in-fun-all-in-earnest little
paper by Jon Postel and Danny Cohen proving mathematically that if there
exist services at protocol layer N there necessarily exist services at
protocol layer N+1, for any positive integer N. TCP is an existence proof
of a service at layer 4, so the hypothesis applies and the inductive
conclusion follows...
Seriously: Of course we have new things such as X, Kerberos, Privacy-
Enhanced Mail, R/A/NFS, and various RPC mechanismms. But there's lots more
needed, such as a good directory service (white pages / yellow pages), a
handy database browser, then one that serious commercial users have been
asking for for over a decade now: a robust transaction protocol, and so on.
But wait; there's more...
> ... I'm inclined to suggest we put our heads together on
> policy and management first ... because the bigger this network gets, the
> more we need new ways to let users use.
End-to-end accountability for service irregularities is I think the biggest
issue. "My ftp just timed out after the first 20 megabytes. Who do I call?"
"Yesterday I could mail to my friend at Kookaburra U.; today it bounces.
Who do I call?" (I'm touting "1 800 THE NREN" as the answer, but I haven't
as yet a clue how to implement the service [IETF WGs notwithstanding].)
