Hold on, before we begin tearing the entire Internet apart, I was referring to the channels
that exist between the subscriber and the head end. Or, in the case of the cable modem
link specifically, between the subscriber's modem and the cable modem termination
system (CMTS) at the serving head end node. Note, the first head end is not necessarily
the primary serving one. Head ends are often clustered, with primaries and secondaries.
And data admin is sometimes handled in some not so intuitive places, due to mergers,
takeouts and consolidations.
And I wasn't referring at this point at least, to the bandwidth pipes on ATHM's wide
area intranet cloud, nor the other upstream carriers' clouds.
There are some 60 or 100+ channels available, depending on the level of upgrade on a
given system, between the head end and the subscriber, to answer one of your questions.
In the future, when the 750 MHz to 1 GHz spectrum opens up for two way and
advanced digital services, many more MPEG and voice services are slated to appear.
The exact channel capacity of this region of the spectrum is not specified, since the mix
of services that each MSO elects to deploy will determine the count at any point in time.
But these are futures. And they are the same ones that have me stumped for the
moment, concerning Greg's questions the other day. This is the region (750 and 1000 MHz) that is slated for both future voice _and_ the "must-carry" digital services, hence the potential conflict, if the must carry load is sufficiently high.
Each of the original lower spectrum channels (between 50 MHz and 500 or 75 MHz, has
a bandwidth of 6 MHz. Any of these, at the discretion of the MSO, could be added to the
group of downstream data channels for use by AOL, and/or the other ISPs.
Those are the channels I'm referring to in the downstream direction. They are regular,
everyday cable TV Channels. Do you suppose that this will cause some confusion at the STB if the user wants to toggle between services?
Upstream allocation is a more complicated and restricted matter, and requires a more
deliberate touch. Upstream would not be as simple as taking a dedicated TV program
channel and redirecting its use to data. I'm not sure just what chunk would be offered up
for AOL's upstream, or how this would be achieved, but between 5 MHz and 40 MHz,
which is known to suffer from power line noise and other forms of ingress, is where the
upstream (and some current cable telephony) services are handled.
I'm sure that the MSO would be able to eke out some quality spectrum for a competitor
to use in this region. <no emoticons necessary>
At the head end itself, there is a high frequency (RF) hand off between the MSO's main
Internet router (say, a typical NT/Bay or Cisco Router) and the fiber portion of the HFC
architecture.
This is where the insertion of the foreign ISP's feeds would take place, at the RF mixer
element. It is at this point that a separate port could be used to attach the router to the
HFC complex, at a different channel frequency, in order to give to the additional cable
tenant, like AOL, separate access to their downstream subscribers. This would take
place bi-directionally, as well.
Looking the other way towards the Internet (or in the case of ATHM, towards their
intranet), the wide area pipe that AOL would use could be a separate T3 line from a
carrier of their own choosing.
Here, the connection is made from the WAN side of the router to their own carrier.
Or they could ride piggy back, for a fee, over ATHM's intranet pipes, and be
dropped off at a peering point or a NAP, somewhere along the way.
These things are doable to only a limited extent, and at a potentially high price for ATHM
down the road. It could lead to ATHM cannibalizing its own structure, if uptake of a
limited potential user group in a given locale favours the interloper.
But someone else said it best earlier. There should only be one physical layer landlord of
the overall infrastructure. The MSO.
And the design criteria behind the HFC model are predicated on the fact that a single
administrator of field units and central site matters would be that same MSO.
Other issues that will arise when these matters get fleshed out, are:
Who administers the upper layer functions of the DOCSIS modem feature set?
It was designed to provide security functions, voip hooks, multimedia capabilities (in
accordance with IMTC and DAVIC specs), QoS settings, etc.
Does the DOCSIS spec get re-written to allow partitioned levels pf control and
administration, depending on who is seizing the line at any point in time? Does this mean
that a subscriber can only subscribe to one service at a time, or will they be owned,
virtually, by the service provider of their first choice?
Do we stop at two potential service providers, or do we include the possibility that any of
the existing 6,500 or so ISPs who want access can control it at will, too?
I don't think so, not even for two, for that would cause violations to all administrative
measures that are implemented for security purposes, that either of the two ISPs would
have implemented for their respective systems and subscribers. Do you think that
security is important to the ISPs? This matter alone almost demands that each ISP has
access to their own channel, separated by guard bands and other difficult to certify
measures.
What happens when there is not a single head end to contend with, rather a cluster of
head ends that are fed by a proprietary fiber ring system with integrated network
management and provisioning? And this provisioning capability is used to remotely
configure modems?
Do both or all ISPs get control of this network management and provisioning tool, for
equal access, simultaneously? Again, I don't think so. NMS systems are tied inextricably
to other operations support systems that contain much guarded information concerning
customers, account information, histories, etc. These would have to be reengineered as
well.
In fact, the more we examine this, the more the word re-engineering pops up, and the
ink on the current iteration of specifications hasn't even dried, nor have those specs even
been fully ratified and implemented, yet.
How is voice over cable administered, and maintained, when the phasing in of this
service requires migration and the potential juggling of spectrum during the tuning
process... when there is a non-involved player sitting in the middle of the frequency
band? Or does AOL want in on the voice aspect too? It would be no more far-fetched
for them to want voice service provisioning capabilities at this point than their desire for
the ability to deliver Internet services. Tell me, why would it be more ridiculous?
Incidentally, this technology is sufficiently diverse and confusing at the same time, that I
went off to find a web site diagram that I could point to as a prop for these explanations.
Suffice it to say that I went to four of sites: Motorola, AS, Next Level/GI, and ADC...
and guess what? Each of them was sufficiently different in nomenclature, style, and
underlying topological profile, that I elected to wing it here, on the fly, trying to be
as generic with my choice of words, as possible. |
