[134] in Back_Bay_LISA
Re: Serial Line connects....
daemon@ATHENA.MIT.EDU (Doug Mildram)
Wed Jan 20 20:03:48 1993
From: Doug Mildram <mildram@xylogics.com>
Date: Wed, 20 Jan 1993 18:57:31 -0500
To: asonnenb@sentry.foxboro.com, bblisa@inset.com, nearnet-tech@nic.near.net
Disclaimer: As sysadmin@xylogics.com, I only use limited SLIP inhouse
and neither SLIP/PPP (YET!) for dialins, because
1) inhouse, all but 1 person wants ethernet speeds for their xtubes.
2) inhouse, I prefer RAM-based xtubes, and slip-booting is a bad idea.
(However, One person is happy using a PROM-based ncd at 38400).
3) terminals and emulators are basically all my users have at home.
But (3) is changing.....PC/Mac slip clients are comin' on strong. One final
word: Use eia (h/w...RTS/CTS) flow control on both ends if you can.-doug
Excerpt from Functional Description of Annex family, R7.0 s/w, 1992
Some may ask "What the heck is ppp?", well the answer is it's like
super SLIP. Like SLIP, it allows a site to use the Annex as WAN
hub, tying distant corners of their IP network together over low
cost phone lines. SLIP is a de facto standard, while emerging PPP
is the IETF de jure standard. And PPP was designed more with
routers in mind, so it is a multi-protocol link which is flexible
and configurable by the two computers. The Annex is no router yet,
but we can still take advantage of PPP. The first implementation
gives the Annex superior dial-in service for workstations and
X-terminals.
The largest and most important thing about PPP is it is built to
change. A massive amount of Internet brain-power is churning out
extensions to PPP. Other high points of PPP and differences from
SLIP include:
PPP can be used on asynchronous or synchronous lines. Many routers
are using it to speak across leased lines at speed ranging from
1200 baud to T1.
PPP uses FCS frame checking, which allows the early detection of
bad packets instead of relying on TCP to detect this much later.
PPP allows each end of the connection to agree on a mutually
acceptable set of features for that connection. This allows for
a much higher degree of "call and play" type connections as
opposed to SLIP where all details must be known at startup.
PPP can run with a much larger MRU than SLIP allowing for economies
of scale. SLIP runs between 256 and 1066 while PPP starts negotiations
at 1500.
As an example Myron Q User (PhD in English) uses his PC at home to
access his work network. The Annex has a PPP port attached to a
modem waiting patiently for work. Myron uses his PC to connect to
the PPP port. The Annex irons out the details of the link, and
most importantly programmaticly authenticates Myron and uses the
sites policy to assign addresses to the end points of the link.
Myron's PC is now on the network, regardless of which Annex port
he's dialed in on, because his address is linked to his name, and
the Annex knows to send any traffic his way that it sees on the
ethernet. Thus the mail daemon, ftp, telnet, or whatever can find
Myron like any other directly connected host. And when Myron exits,
the modem is available for the next user. This scenario work best
with the endpoint addresses being all the same network.
FUTURES: (since annex doesnt by itself do dialback at this time:)
But this is just the tip of the internetworking iceberg. As Annex
evolution moves it towards routing, PPP will be there to build
upon. Dialup Appletalk would rely on PPP for its link-layer. A
dialout capability combined with active routing would give us
Netblazer functionality. (and so on. Hope you enjoyed this .----doug)
GORY DETAILS: rfc 1171, 1172, 1331.
--
Mail for the `bblisa' list goes to `bblisa@inset.com'.
ONLY mailing list requests go to `bblisa-request@inset.com'!
bblisa administration messages go to `bblisa-admin@inset.com'.
