[comp.protocols.tcp-ip] SUMMARY Digital phones for SLIP etc

efb@suned1.Nswses.Navy.MIL (Everett F Batey) (03/10/91)

Thank you to the many who responded.  Hope the edits do proper justice to
all.  This is less filtered and sorted than I hoped for BUT feed this
to the printer and get out a six-pack.  Keep or edit out form feeds
each +/- 58 lines.  AN elegant short haul alternative at bottom.

My goals are to get a few more USN sites glued together better and ultimately
join other DoN sites in trying to get Navy sites inter-connected by common
resources and get out of the unique solutions rut.

The following is a personal work and only attempts to collect the opinions
of others.  You are welcome to this .. IN THESE TERMS .. Not an OFFICIAL
Statement.  


On Mar 2, 19:51, cliff bedore* wrote:
 I have seen ads in Black Box and other places adds for "short haul" modems
 that do what you want.  I seem to remember that they are not terribly
 expensive ($100-$200?).  I'm not at work but if you don't have any luck
 elsewhere, send me mail and I'll look it up on Monday.  I've been trying to
 get SLIP running between XENIX systems over USR HST modems and they won't
 work at over 2400 baud.  Slip itself runs just fine at 19.2 but these modems
 won't cut it.  Response at 9600 baud is reasonable for 1-2 logins so if you
 can get one of these long haul modems to work, that should be fast enough. 
}-- End of excerpt from cliff bedore*

On Mar 3,  7:46, John Scoggin wrote:
 We have some Northern Telecom Async-Sync Interface Modules, and they are very
 reliable up to 56 KBPS!  So long as your transmission facilities are clean (we
 use our own fiber cable to provide T-1 service between switches), you should
 have absolutely no problem.  I believe that the async will run up to 19.2 KBPS.
 You need to be careful with the Suns, though; I have read several comments
 about over-running the async ports on them when using SLIP. (Haven't tried it
 myself)  Email: scoggin@delmarva.com
}-- End of excerpt from John Scoggin

On Mar 3, 14:42, Wayne Sung wrote:
 I'll venture answers to two of your possibilities:
 1) if you have direct wire pairs line drivers can work quite well. Which
 exact kind to use depends on whether the lines are loaded. If they are, you
 will need voice band drivers which are less desirable than the Telebit
 possibility. If they are not loaded you can even get 56k at 3 miles. I like
 using DDS CSU/DSU's for line drivers because they are more tolerant of line
 problems.
 2) using any two Telebits at 9600 PEP (and in fact needing only one pair
 instead of two) you should use the leased line mode. This way they can
 restart themselves. We have a number of dedicated control circuits which
 are done this way (although over microwave voice channels). You also get
 error handling which a simple line driver will not give you.
}-- End of excerpt from Wayne Sung

On Mar 3, 17:11, Robert M. Enger wrote:
 If you have access to a 4-wire metalic facility,
 why not go for the gusto.  You should be
 able to run atleast 56Kbps straight end to end,
 with QUALITY csu/dsu equipment.
 You MAY be able to run T1 straight end to end.
 If not, you can certainly get away with it if
 you can put a T1 repeater somewhere along the way.
 (Check with manufacturers of the T1 csu/dsu 
 equipment; you need good HOT line drivers, etc)
}-- End of excerpt from Robert M. Enger

On Mar 6, 14:21, elroy!ISI.EDU!prue wrote:
} Subject: Re:  Digital phones for SLIP circuits, 56KB tariff data, too
   Both GTE and PacBell use an ADN intra-LATA tariff that provides
   digital dedicated service for ~$100 +$6/mile per mo.  The interesting thing
  .. thank you very much .. is that substantially what LOS NETTOS is 
  built upon ?   What class name is that service ??
 
 Los Nettos uses T1 service for all regular members.  T1 runs at 1.544 Mb/s
 about 27 times the 56K service I mentioned.  Los Nettos Associates connect
 to member networks at whatever rate they want to pay for.  The slowest is 9600
 but most use 56K ADN (Advanced Digital Network) service.  Some Associates

 use T1 service too.
}-- End of excerpt from elroy!ISI.EDU!prue

On Mar 6, 11:12, elroy!ISI.EDU!prue wrote:
 I dont have specific knowledge of SLIP but I dont think that it depends on
 control signals for flow control.  If your medium is full duplex as are digital
 services that I know of and the modern dial modems, I can not imagine that
 you would have problems.  Some link protocols have flow control based on
 X-ON and X-OFF.  This messes up transparency.  Others use control leads of
 the RS232 interface that would be seen at the other end.  This sort of flow 
 control is often used when the two devices are cable lengths apart not dial
 modem lengths apart.  I don't think SLIP does flow control.
 
 You mentioned using Northern Telecom 9600 digital service.  I know nothing
 of this type service but an interesting thing to note is the ADN tariff
 we are using.  Both GTE and PacBell use an ADN intra-LATA tariff that provides
 digital dedicated service for ~$100 +$6/mile per month.  The interesting thing
 is that the rate is the same for 2400 or 56000 bps.
}-- End of excerpt from elroy!ISI.EDU!prue

On Mar 3, 23:03, efb TO: Wayne Sung wrote:
 If I read you correctly, the telebit, as delivered with RJ11s?, can also run
 on simple metallic pairs with no CO battery, etc or however metallic lines
 are delivered.  IS THIS THE CASE ?  Just a reconfig of the inerts of the 
 telebit ???  /Ev/
}-- End of excerpt from TO: Wayne Sung

On Mar 4, 21:27, TO: Bob Sutterfield wrote:
 Bob .. thanks .. the PPP sounds like the better solution .. except .. on
 the remote end is a MVAX-II running Berkeley-TAHOE, OR a Sun3/260 running
 SunOS 4.1.1.  Noticed much SPARC Sun stuff rans on the S-3s at 4.1.1.
 
 Do you have reason to think the PPP code might make the grade on Sun 3s with
 only trivial fixing ???  Thank you very much .. /Ev/
 
 On Mar 4, 10:24, Bob Sutterfield wrote:
     money for modems, we are trying to pick the most trustworthy and
     easy to accomplish SLIP link.  Sites are within 3 miles, wire, 
  I think you are making your specification too specific.  Do you really
  need to run RFC1055 SLIP, or is the need for an IP channel over a
 
  PPP, the newer RFC1171 Point to Point Protocol.  In fact, we have both
  PPP and SLIP installed in the same kernels and they both work fine,
 
  Whichever you use, be sure it implements RFC1144 TCP header
  compression.  It will improve your performance remarkably.
 
  Get ftp.ee.lbl.gov:cslipbeta.tar.Z (if you must run SLIP) or
  tut.cis.ohio-state.edu:pub/ppp/ppp-sparc4.1.tar.Z (if you want a
  reliable, flexible, high-performance serial line IP connection instead
  of SLIP).
     - We have some available Telebit modems, 
  SPARCstations connected by a T2500 and a TB+, we see 1.3-1.5
  Kbytes/sec FTP throughput, one-way.  Interactive response is
  survivable.  

 
     Is there a good way to use these locally over an analog delivered
     voice grade phone line ?
 
  In the Telebit Trailblazer Plus reference manual, please see Appendix
  F, "Leased Line Considerations".  In the manuals for other modem
  models, the appendix may be numbered differently.  
 
     - How would YOU rate from YOUR PERSONAL experience these options,
     for cost effectiveness 
 
  DEFICIENCIES section of RFC1055 and the Introduction of RFC1171 for a
  discussion of why you should avoid SLIP unless you absolutely must use
 
  Our company will soon release a product that implements both SLIP and
  PPP in a convenient-to-install-and-use package.  We have no explicit
  plans right now to support the Sun-3 line, but it should be no great
  problem to do the port if you need it.  Perhaps we could have
  temporary Internet access to your systems while we do the work.
  Please contact Jamey Laskey at +1-800-558-7827 or write to
  marketing@morningstar.com to find out more.
 }-- End of excerpt from Bob Sutterfield
}-- End of excerpt from TO: Bob Sutterfield

On Mar 4, 14:12, Robert M. Enger wrote:
 We run Dowty T1 csu/dsu units.  they are model 7055060
 
 I recall they cost about 1500 each.
 These are extended super frame units (which
 allow out of band end-to-end monitoring
 of link error rate, remote monitoring of
 distant end csu/dsu unit, etc).
 
 They are also capable of B8ZS operation
 allowing you to put full 1.536Mbps user data.
 They act as the dsu (putting on framing, etc)
 so that you can plug the sun, or a router 
 directly into this unit (v.35 connector, 
 need v.35 to rs422 adapter from Blackbox if
 computer has rs422 connector).
 
 You won't be able to drive these from the console
 port on a sun (and you wouldn't want to anyway
 since console port is NOT DMA and causes an
 interrupt per character :-(   ).
 You'll need to get a Synchronous data board
 for your suns, or use routers.  The routers
 will remove some load from your suns.
 
 I like Cisco AGS+ routers, but they will be
 overkill for your situation (probably, don't
 know all your specifics).
}-- End of excerpt from Robert M. Enger

On Mar 5,  9:57, Bob Sutterfield wrote:
    Bob .. thanks .. the PPP sounds like the better solution .. except
    .. on the remote end is a MVAX-II running Berkeley-TAHOE, OR a
    Sun3/260 running SunOS 4.1.1.

 
 Get PPP for VAX BSD4.3 from lancaster.andrew.cmu.edu:pub/ppp/ppp.shar.
 This is Drew Perkins' <ddp@andrew.cmu.edu> original PPP
 implementation, developed (as I understand things) as a proof exercise
 while writing the RFCs.  It's the ancestor of the stuff in
 tut.cis.ohio-state.edu:pub/ppp/ppp-sparc4.1.tar.Z.
 
    Noticed much SPARC Sun stuff rans on the S-3s at 4.1.1.  Do you
    have reason to think the PPP code might make the grade on Sun 3s
    with only trivial fixing ???
 
 Yes, and I think folks are already doing just that.  Brad Clements
 <bkc@omnigate.clarkson.edu> took Drew Perkins' VAX code and ported it
 to a Sun386i under SunOS 4.0.1, adding TCP header compression and
 STREAMS support along the way.  Karl Fox <karl@morningstar.com> took
 that result and ported it to a Sun4 under 4.0.3, 4.0.3c, 4.1, and
 4.1.1, removing some assumptions and cleaning up some bugs along the
 way.  I strongly suspect that you'll be able to build the same 4.1
 code into your Motorola 4.1 kernel with no changes at all.  If you
 find any assumptions that remain after the Intel-to-SPARC port that
 make it unable to run on a Motorola machine, please let Karl and me
 know and we'll plow them back into the distribution again.  I doubt
 there are any such problems, and I think I've heard of people running
 this stuff on their Sun3s, but I can't put my finger on the reference
 just now.
}-- End of excerpt from Bob Sutterfield

On Mar 6, 13:23, TO: elroy!ISI.EDU!prue wrote:
} Subject: Re:  Digital phones for SLIP circuits, 56KB tariff data, too
  You mentioned using Northern Telecom 9600 digital service.  
   I mis-spoke .. feature of Meridian Phones .. digital with analog cards at
   subscriber end.
  of this type service but an interesting thing to note is the ADN tariff
  we are using.  Both GTE and PacBell use an ADN intra-LATA tariff that provides
  digital dedicated service for ~$100 +$6/mile per month.  The interesting thing
 .. thank you very much .. is that substantially what LOS NETTOS is 
 built upon ?   What class name is that service ??
  is that the rate is the same for 2400 or 56000 bps.
}-- End of excerpt from TO: elroy!ISI.EDU!prue

On Mar 9,  7:59, C Pratt wrote: } Subject: SLIP on TP ( Ed. Twisted pair ? )
     If your pairs are dry, you can implement full 802.3 over 2 pair for 2300
 per end with Network Application Tech Bridges and Black Box or RAD 64KB line
 drivers. These are good for 6 miles (422/449) and WORK. This is how we did the
 floors and buildings in CC after all the high tech solutions failed (due to
 artificial roadblocks set up by NC building managers and the phone company
 (C&P). The pairs we had to use are noisiest imaginable and elevator motors
 induced massive noise spikes on the line every time someone went upstairs. I
 had no money left as my predecessors spent it all on equipment that would not
 work. We stuck serial connections on all the systems just to get something
 going and then one of my vendors came up with these "cheap" bridges. 64KB isnt
 T1 but T1 is pricey. The bridges will standalone or you can buy even cheaper
 versions that do the same thing but are just cards that fit in a dead AT (they
 don't use the AT buss for anything but power). This makes a cheap multiport
 bridge for the headend of a star.
}-- End of excerpt from C Pratt
-- 
 +  efb@suned1.nswses.Navy.MIL  efb@gcpacix.uucp  efb@gcpacix.cotdazr.org +
 +  efb@nosc.mil   WA6CRE    Gold Coast Sun Users   Vta-SB-SLO DECUS  gnu +
 +  Opinions, MINE, NOT Uncle Sam_s | b-news postmaster xntp dns  WAFFLE  +