Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!snorkelwacker!spdcc!dyer
From: dyer@spdcc.COM (Steve Dyer)
Newsgroups: comp.protocols.tcp-ip
Subject: Re: How is TCP/IP carried on T1's?
Message-ID: <1095@ursa-major.SPDCC.COM>
Date: 4 Jan 90 00:26:04 GMT
References: <2695@ucsfcca.ucsf.edu>
Reply-To: dyer@ursa-major.spdcc.COM (Steve Dyer)
Organization: S.P. Dyer Computer Consulting, Cambridge MA
Lines: 28

In article <2695@ucsfcca.ucsf.edu> jst@cca.ucsf.edu (Joe Stong) writes:
>Someone I know claims that most TCP/IP packets transmitted on T1 lines are
>imbedded in X.25 protocol packets.  Is this true, or what is actually
>used?   Or it TCP/IP alone with maybe an alternate checksum (CRC) sufficient?

T1 is just a bit pipe.  You usually have some sort of synchronous
hardware/firmware on each end which also provides some form of point-to-
point protocol.  This could be byte-oriented DDCMP, as with DEC DMR-11
sync interfaces, or the bit-oriented LAPB/HDLC, sometimes referred to
as X.25 level II, or even a minimal encapsulation with checksum, as with
the DEC DMR-11 in "maintenance mode".  What's important is that the two
ends agree.  Proprietary routers have already made their choices.
I guess in the future, some form of the upcoming Point-to-Point Protocol
could be used.

I don't have any data on the reliability of data streams over T1 lines,
which might drive a selection of protocol.  We routinely ran in maintenance
mode with DMR-11's between Harvard and MIT in the early days of the JVNCnet.
The hardware discarded packets with bad link-level checksums, but otherwise
did not perform any retransmissions.  The sentiment was that it was undesirable
to have lossiness be transformed into delay in the calculation of TCP
round trip time and retransmission timeouts.


-- 
Steve Dyer
dyer@ursa-major.spdcc.com aka {ima,harvard,rayssd,linus,m2c}!spdcc!dyer
dyer@arktouros.mit.edu, dyer@hstbme.mit.edu