Received: from eff.org by kragar.eff.org with SMTP id AA19129
(5.65c/IDA-1.4.4 for <
[email protected]>);
Mon, 13 Apr 1992 11:39:34 -0400
Received: by eff.org id AA18534
(5.65c/IDA-1.4.4 for pub-infra-exploder); Mon, 13 Apr 1992 11:18:20 -0400
Date: Mon, 13 Apr 1992 11:18:17 -0400
Message-Id: <
[email protected]>
From: <
[email protected]>
Subject: Frankston on ISDN
To:
[email protected] (pub-infra mailing list)
Prescript: I've been putting these comments together for a while, but at
some
point, I've got to ship the product even if it is not perfect nor complete.
With that caveat in mind, I'm submitting this commentary on ISDN and the
rest
of the universe.
I've just read through D.P.U. 91-63-B of the Commonwealth of Massachusetts
Department of Public utilities which is the response to the ISDN filings.
I've already Prodigy's briefs on the subject. Though the language was a
bit
long winded, it was, to my surprise, fascinating. I don't know how these
compare to what the situation is in other states, but there is some
spirited
opposition to NET's attempt to sell ISDN services at a high rate. The
Prodigy filing also incorporated some of Mitch Kapor's research.
I should also strongly emphasize that I am not a lawyer, I'm not well
versed
in regulatory law. I'm more a dabbler/kibbitzer in these matters, then an
expert. Given all that, I'll attempt to give a biased summary of the
filings
and my reactions. More my reactions than summary.
The basic position of NET is that ISDN is an optional offering that should
be
priced at competitive rates. Where competitive means "what the market will
bear". Prodigy's position is that ISDN should be a basic part of the
communications infrastructure and should be priced at a rate that would
encourage its use, or at least, not discriminate against its use. NET
referred to Prodigy's view as a "field of dreams" wherein Prodigy expects
that if ISDN were available, people would flock to it. NET claimed its
studies showed that there are people who would pay high rates for ISDN and
thus it should be priced for the known customers. I call this "railroad
pricing" referring to the days when the railroads were in decline and kept
increasing their rates to get the same return from fewer passengers and
thus
reduced the number of passengers etc. Even worse, for communications
offering, providing only one hand so that people can experiment with
clapping
has its limits.
There is much discussion on what the actual costs of ISDN deployment are.
This gets complicated because the costs of ISDN components vary due to
accounting considerations as well as purchase price variations due to one
time offers, quantity discounts and startup costs. The distinction between
hardware, software and other components is not clear, so I resort to the
technical term "stuff". The fact that these are not broken out makes it
hard
to evaluate some of the claims.
The DPU seems to be caught in the middle. It seems to buy into the
infrastructure argument but is very conservative on limiting NETs rates
since
NET must be allowed to recover its costs.
Now a word from our sponsor -- me. I had a number of disparate (knowing
how
bad spelling is on the net, I should point out that that is not a typo for
desperate) reactions to reading these filings.
One question is whether ISDN is the right service for data. Some of the
DPU
discussion was on the relationship of ISDN as a data transport with
switched
56KB (an example of high priced service). But there was no discussion on
how
to provide a connectivity other than point to point bulk data transfer. I
realize the advantage of concentrating on ISDN issues is that there is at
least some agreement on what ISDN is -- a necessary prerequisite for
rational
debate. Given the grief that NET is giving over ISDN deployment, asking
for
really different services seems to be an uphill battle.
Which brings us to NET as a consumer buying merchandise off the shelf. It
seems that once they've bought into an exchange (often the DMS-100) they
are
captives of their maintenance plan with the exchange. I can't imagine them
buying anything nontrivial from anyone but Northern-Telecom for its DMS-100
COs. It doesn't seem they have much negotiating leverage. It would be
nice
to see the CO become a more distributed entity that allowed more mixing and
matching from different vendors. For now, at least, it is not clear how to
build such a system. This further concentrates debate on off-the-shelf
ISDN
because alternatives are problematic.
NET whines about the difficulty of providing ISDN, yet they seem to have no
problems if you want Intellipath and Centrex, both of which are ISDN-based
services. This seems to be far from a level playing field. I'd like to
see
a situation in which NET couldn't base any of its services on ISDN unless
others gain the same access. Of course, the fact that the ISDN services
seem
to run in the same switch as ISDN itself means that they can provide the
services without the complexities of providing ISDN to third parties. This
goes back to the issue of the monolithic CO. Perhaps ISDN can be a
mechanism
for brining CO capabilities outside the physical (or even logical) machine
so
that the protocols necessary for these services are provided at arms length
and thus provide a mechanism for a marketplace.
This issue of ISDN as a set of protocols for implementing a marketplace is
an
interesting one but not fully fleshed out in the current ISDN protocols.
It
should receive explicit attention in its own right. Given all this, it
still
galls me to see NET refer to capabilities by their service product names
rather than the generic features. It also recalls the problems that "good'
COCOTS have in trying to compete with NET in the pay phone business since
NET
gives itself a great deal on the costs of phone calls (of course, the fact
that Massachusetts still mandates $.10 for NET pay phones is probably also
a
factor -- something that bespeaks a strong DPU which might make the state a
good place for ISDN advocacy).
The cellular phone network offers an interesting case study. In following
the discussions of features of the cellular phone network, it feels like
amateur night wherein features are cobbled together by kludging together
disparate systems. A lot of the feature set depends on whether one happens
to have an Ericsson or Motorola switch and what sort of jumpers have been
placed between them. I get the impression that many features are
implemented
by placing a PC (personal computer) offnet and having it send back DTMF
codes. I see this a symptom of the complexity of making any changes in the
network. At least, in the network as currently architected
The ISDN and cellular problems illustrate the problem of what happens when
one buys a complete service from a monopoly. Unlike the PC world of mix
and
match, you get all or nothing. A long term agenda should be to go beyond
simply breaking of ATT to architecting a communications infrastructure that
consists of components. This is very very difficult, especially when
coupled
with requirements for reliable and predictable service. It is less clear
that the current approach is more amenable to graceful evolution.
A more modest approach is to encourage competition on the local loop.
Perhaps the RBOCs should be divested of their control of the right of way
and
instead, all parties would have equal access to the communications right of
way at a physical level. Access at a logical or signal level is more
complicated. We'll see some of this in action now that Cable companies are
becoming more of a force for loop competition. Cable company monopolies
are
anther topic I won't get into here.
Back to my Ox. The current network is designed for voice communications
with
services like switched 56KB being viewed as expensive premium services.
But
the reality is just the opposite. Voice is very demanding of the network
whereas data communications is very forgiving of delays and can recover
from
errors (OK, voice can tolerate many errors that give data fits). So why is
voice cheap and data expensive? There are some answers in the current
network architecture but these are not intrinsic.
The other aspect is the circuit switched model for data communications.
Admittedly it is possible to get an X.25 connection that does provide a
switched service but I'm not confident that it is sufficiently standard for
me to assume I can make a very cheap quick connection to a service and be
charged accordingly. If I want to get one stock price, how much overhead
is
involved? If I want to keep simultaneous connections to multiple services
is
there a holding charge? I realize that there is a contradiction between my
asking for a raw service from telco and the ability to then buy enhanced
services from other parties. But is the raw service copper to the CO and a
voice path or is it a datapath. If a third party provides the packet
service, do all messages have to travel through the network and then get
redispatched? Are there sufficient standards for things to "just work"?
This brings us to the concept of intra CO tariffs These do exist for
Centrex
and might exist for early ISDN capabilities which can be supported within a
switch but which must await protocol upgrades (SS7?) in order to
communicate
with other exchanges. I can image that a call within the exchange being
essentially free but having a significant charge to call the next town. Or
should social policy minimize this? We already have the example of cable
TV
systems where I simply cannot get broadcasts from the next town -- a very
bad
precedent but something we accept as if it were natural rather than a
kludge
while we await BISDN (where BISDN is a code word for switched video but not
limited to ISDN protocols).
Postscript. I've seen mention that NET has adopted ISDN pricing that is
akin
to standard message unit service but haven't seen the details. I've also
received a brochure from Nynex touting Basic Exchange Service which seems
to
be their ISDN Centrex replacement that lists a menu of features that you
can
select 3 of plus optional features. Of course, it is these services that
are
being offered, not "raw" ISDN.
[Pricing in Mass. is 2.6 cents per minute for the first minute for
residential service, 1.6 cents per minute thereafter. For businessis, the
rates are 9.6 cents for the initial minute and 1.6 cents thereafter. This
covers service to the CO. Inter-office tariffs will be filed by the end of
the year. - Mitch Kapor]
