NetNews Usenet Archive 1992 #31/NN_1992_31.iso/spool/comp/dcom/isdn/1034

Text File | 1992-12-23 | 3.9 KB | 73 lines

Newsgroups: comp.dcom.isdn
Path: sparky!uunet!uunet.ca!xenitec!vances
From: [e-mail] (Vance Shipley)
Subject: Asynchronous TAs
Organization: SwitchView Inc., Waterloo, Ontario
Date: Wed, 23 Dec 1992 19:27:10 GMT
Message-ID: <[e-mail]>
Summary: where are they?
Keywords: RS-232 asynch synch 57.6K 56K
Lines: 61


In order for ISDN to gain acceptance in the marketplace it must provide
something of value which is unavailable with POTS technology.  The issue
of ISDN->POTS conectivity has been discussed here.  Providing a modem
pool in the CO is one solution, terminal adaptors which provide the
conversion is another (I believe Canon have a G4 fax which will also
receive G3 transmissions from POTS machines).  Here is my pet peeve.

Of all the ISDN terminal adaptors I have seen very few provide asynchronous
speeds above 19.2K.  None of the TAs I have seen provide faster than 38.4K.
Of course they all provide 56 or 64K synchronous.  I'm sure this makes
sense when viewed from the position of evolving the current telecommunications
solutions to ISDN.  Currently modems provide asynchronous interfaces at
speeds up to 19.2K and sometimes 38.4K while DDS provides leased 56K
synchronous connections.  The problem here is that by maintaining this
limited view the TA manufacturers have ensured that they will cater solely
to the high end users.

For the billion PC users out there without synchronous cards and software
there is no benefit in getting ISDN.  These people have, or could get, V.42bis
modems which they believe will get them 19.2K throughput.  If they get an ISDN
terminal adaptor they could achieve a real 19.2K conection.  There would be
a gain here if you could explain the issue of compression.  If they were to
get a TA which gave them 38.4K they would have a better than two fold gain.
This is just enough to make it worth the trouble, maybe.

A TA which would provide a 57.6K asynchronous connection would give a three
fold increase over there expectation of 19.2K.  An asynchronous speed of 76.8K
would provide a 400% percent increase over the expected throughput of POTS.
The actual throughput, before compression is more like 14.4K at best on POTS.
That would translate into a 533% increase at 76.8K.  Comparing this to the
more common throughput of 9600 baud would yield an 800% increase at 78.6K.

Once the start/stop/idle bits of an asynchronous byte stream are removed it
can be placed through a synchronous channel of a much lower speed.  There is
at least one company making a synch->asynch convertor which does just this.
At this years CBTA show I found a Sun 386i running a demonstration in
Northern Telecom's booth.  It was running some sort of multimedia over a
switched 56K Datapath setup.  Sitting atop the TA was a convertor which
took 57.6K asynch from the Sun 386i and fed it to the synch only TA at 56K.

There is a popular misconception that asynch can only support 19.2K.  This
is simply untrue, any lowly 16540 UART can handle 57.6, 76.8 and even 115.2K!
This means that almost all PCs in use today can handle the speeds I suggest.
The cost of doing synchronous on a PC today is astronomical in comparison.
Of course the other choice here is to get a TA which goes in the PC and
eliminates the asynchronous problem.  This is only a solution for DOS users
because of driver availability.

I read an article recently about a group of astronomers who were working at
home with X terminals over ISDN.  Because the X terminals only provided an
asynchronous serial interface (in addition to the ethernet) they were only
using 38.4K.  The article was giving this as an example of how ISDN allowed
users to do things never before practical.  While this was essentially true
to me it was a great example of what is wrong with the cuurent state of
affairs in the ISDN world.


--
Vance Shipley
[e-mail]         [e-mail]         [e-mail]