Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 19 Mar 1996 10:52:02 -0600
From:      Tony Kimball <alk@Think.COM>
To:        lehey.pad@sni.de
Cc:        hackers@freebsd.org
Subject:   Re: ADSL
Message-ID:  <199603191652.KAA10764@compound>
In-Reply-To: <199603191609.RAA20719@nixpbe.pdb.sni.de> (message from Greg Lehey on Tue, 19 Mar 96 17:07:17 MET)

next in thread | previous in thread | raw e-mail | index | archive | help
   I consider it extremely unlikely
   that it will find much following in mainland Europe.

I'll emit an evil laugh as I watch Europe sink into the stone age:)

   > No, a factor of 10-80 in performance, the opportunity for real
   > videoconferencing and VOD is just too much weight for ISDN to bear.

   I don't see that factor out there.  The Web page you point me to talks
   about downlink speeds between 384 kb/s and 4 Mb/s, and uplink speeds
   between 64 kb/s and "10 times that much".  That's hardly any faster
   than regular ISDN.

Downlink speeds dominate.  If you want symmetric, go HDSL not ASDL.
That's not mass-market technology, though, so while complimentary
development of ADSL components will reduce the cost of HDSL equipment,
it will never get to the raw commodity prices that the 100M unit 
volumes of ADSL modems will achieve.  (Alternatively, use an MPPP
bridge/router with reverse-oriented ADSL modem pairs.)

Therefore, I only care about downlink speeds, in arguing obsolescence.
(Although for passable videoconferencing I would require at least 384k
uplink -- still, I don't expect much of that until a future generation
of products.  No, VOD is the killer app here.)

384k downlink speeds are in the far end of the service range.  Typical
application speeds will vary from 1.5 to 6.  In short-haul loops, VDSL
will peak out at ~50Mbps, but will be too pricey, again.  In those
short-haul loops, ADSL can run to 8, using proprietary vendor
encodings.  It is from the latter (best-case) figure that I derive the
upper end of my multiplier.  To be fair I should have said "3-80"
times instead of "10-80".  90% of the sample will fit in the "10-40"
range.  It's still far too big a factor.  The infrastructure costs are
*the*same*as*ISDN*.  And *it*runs*faster*.  And *it*supports* TV*.
Computers are nothing.  TVs are everything.

   On top of that, there are a lot of disadvantages that I can see: from
   the Web page, it would seem as if every "modem" manufacturer is doing
   his own thing.  

There are basically three encodings in production, and each
interoperates with other vendor equipment in the same class, in
theory.  Practice in general is too early to close on, but certainly
that CAP modems (Willtel and AT&T) do interoperate is abundantly well
demonstrated.  As an end-user, I don't really care, though, because
all I want is to interoperate with my Central Information Service
Provider.

   This means that I might be able to set up a point to
   point link with one ISP, but doesn's seem to be a way to connect to
   multiple, unrelated sites.  ISDN *does* give you that advantage, even
   today.

Your ISP will provide the link between ISDN/POTS and the ADSL line.
Think like an end-user: You do not care about switching and routing
and uch.  All you want is a fat pipe.  Let the pros take care of
switching.  ISP==Telco==VideoLibrary==Cable==DSS.  One bill to pay
each month.

   > PairGain modems come out in June, and the Willtel CAP offering comes
   > out in October.  You may argue about the relative obsolescence of
   > ISDN today, but come Christmas time, any further arguments will be
   > silly.

   OK, let's have a bet about it.

I don't bet.  I invest.  We'll see who pays the most capital gains tax:)









Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199603191652.KAA10764>