Dave, while your problems are basically personal bandwidth related, mine are fair weighted
Ms.-queues. I'm being accused by some on the home front, rightfully so no contest, of not
allocating my personal resources in a fair way, in other words, and to that I must agree and
adjust.
It's sorta like a debit based scheme for qos, when you find yourself enjoying yourself too
much at the expense of ignoring others. So, I'll be brief here tonight (at least until everyone
else goes to sleep), and return to these points when my allowable time accrues (now we're
dealing with 'credit' based qos) to the point where I can do your comments more justice.
For now, however, I'd like to simply point out that providing Layers 1 thru 4 (i.e., physical
(optical), data link (hybrid 10GbE/ATM?), network (IP) and transport (TCP)), as would
any ISP (plus the normal pop/web services, and some caching, perhaps) would be a full
time job to start off with, and to go deep into the future with, without the need for bringing
in whole other industries, which are content and programing. The way I see it, the platform
could accommodate all comers in those fields, for a fee. And have plenty of room left over,
if designed right. (Hmmm.. where have I heard that before?)
If we look at ATHM, being the closest thing to a residential cable modem ISP that I can
think of (although they are being held back radically from the kind of delivery schemes
we've been proposing here, due to their DOCSIS orientation). We could analyze what they've
been through in a more general sense and form some good opinions, I think. What to do,
and what not to do as we cogitate over what this thing should look like. I wont do that now,
because of time, but maybe this is an angle we should examine further.
Stated another way, what would Milo now do differently if he had
it to do over again, and how would having limitless access to fiber
to the home affect his strategy?
Right now, I think that HOME would do wonders for itself if it rounded out its SP 1-->4
features and services, and left content to the other industry. Then again, HOME is an ISP,
and they have their ownership issues, etc., and T now appears to be coming around to
admitting that it want them in total. But aside from these issues, examining their recent past vis
a vis their intranet design platform, makes for a good study.
If we limit our investigation to only one aspect of their services it should serve to
demonstrate that 1 thru 4 is not all the no-brainer that many folks would ascribe to it.
A real close look at many ISPs, in fact, demonstrates that there are tons of problems with
bad and broken architecture that need to be addressed within the confines of 1-->4, if it is
going to be "real" Internet, as opposed to quasi Internet.
By Quasi I mean ad-hoc'ed kludges which have come about through growing pains, and
stresses caused by commercialization before the architects had a chance to prepare for the
deluge the 'Net has seen over the past six years.
An example: Quasi is sometimes characterized by the almost universal use of network
address translation (NAT), DHCP, etc.. where end users never wind up with the same
address twice. Very big problem which threatens to minimize the true potential of the 'Net.
This isn't the way the Internet was intended to work, in other words. There is a whole story
behind this stemming from address shortages, the need to reconcile at some point the issues
surrounding IP V.4 vs. V.6 and the problems stemming therefrom, etc., that we can get into
some other time. But why is this important?
Keepping in mind that I only mentioned NAT as one example... I would like to state that it
is important because if you don't have an IP address that identifies "you," uniquely,
_at_all_times_, then you don't have end to end Internet, which means that you have
something else: Quasi Internet. You need a universal identifier (an IP address of your own)
for true IP voice (where the client sits on your desktop, not in the central office), for remote
print services when you are on the road, say, or for just about anything that would require
finding "you" at all times, without using application layer gateways (ALGs) written
specifically for every application you intend to use.
Okay, there are work arounds for NAT, and I mentioned one of them that will work
)sometimes(, ALGs. But that's not the point. The point is that the next generation of cable
tv operator, if we are to believe our own drummings, will be characterized by very
sophisticated ISPs, and not dirt to anything players.
In fact, our hypothetical model would do very well to outsource the entire cable laying
operation, which is merely another horizontal function that could be handled at a business
level to outside contractors, and not have anything at all to do with it. And there are other
areas that could be outsourced, as well. I wont take this too far here, but just about every
component of the next gen cable ISP could be outsourced, if you think about it, and still not
get involved with content.
And why not get involved with content? Because if the next gen FTTH architecture is truly as
robust, capacious and dynamically resilient as we profess, then it should be able to
accommodate multiple content providers, and there will be plenty of content providers as
the 'Net matures. Why be bogged down with a vertically integrated operation which caters
to only one? One company, one consortium, one modality, one anything? As a "true"
broadband platform capable of delivering all services in digital format (as opposed to today's
analog r-f schemes which are supported by CableLabs designs, space should be allocable to
all.
For a price.
----
I'll get back to these points tomorrow sometime. I just didn't want your excellent review of
Clearworks, and your other comments about dirt to data to go unacknowledged.
FAC |
