You are right, of course, re-segmenting would include the downstream, as well. It didn't occur to me to mention that because that is the primary implied reason for doing the re-segmenting, in the first place.
The whole idea is to dramatically reduce the level of contention per channel and this is done through the creation and isolation of smaller groups of users on the HFC portion of the architecture than would exist on the larger universe of users that currently exists as a group, in total. This is very much like the principle of having smaller class sizes in school, where the teacher (head end resources) could devote more attention to each pupil (the cable modem attached device).
Both in the downstream, as well as the upstream... although it could be in the downstream only, or disproportionately more in the downstream, if the cable operator so deems.
I would appreciate at this point if David Horne wouldn't mind speaking up on this matter, since these are arbitrary decisions, and I think that David has a better feel for what operators are actually doing in the field, in these regards.
I get the sense that the MSOs want to improve delivery to the residence (which is a mandatory) but not necessarily bolster the upstream to symmetrical proportions, although there is a need to do something in the upstream, as well, to be sure. Some will create symmetry, no doubt, but I'm not so sure that all will even come close to symmetry.
I think that they will continue to look at their model as being an asymmetric one, except where so/ho and telecommutes are concerned, which then may require something on the order of a CMTO- or TERN-like modem approach in order to achieve symmetry, or increased upstream capabilities, on an individual case basis. This is the dreaded tiered service, which assures greater levels of bandwidth and QoS, that we speak of from time to time.
"Lastly, I hate to ask an imprecise question to a person so determined to be precise but I will anyway: How many homes can the current upstream support in a single node? "It depends", Frank replies. I hear the answer already. "
Thanks, Eric.. I needed a good laugh today.
"What I'm looking for is a sense of things. Right now it is ~500 homes sharing a "wire"."
I have heard AHhaha mention this figure several times, and it is commonly used as a default by others. But the fact is that there could be anywhere from a couple of hundred to 2,000 users or more (probably more in some of the earlier rebuilds, and those which didn't plan on Moores Law being so accurate) for cable TV services.
Consider, you can do a construction of 5000 homes passed, expecting an uptake of only 30% or 40%, which would top out at 2,000 viewers if expectations were accurate. What happens, though, when the uptake reaches 50 or 60%?
The proper thing to do would be to run shunts across some part of it, to break it down to 1500 homes or fewer, apiece, or something similar to that. But, as we all know, things are never that simple or cut and dried, requiring engineering reviews, additional opto-electronics/dwdms, possibly additional amplifiers, and so on.
Odds are even up that they don't get tended to in an immediate time frame, and linger on into the ages. That is, of course, until an exigency like cable modem congestion occurs, and something has to be done pronto (which could take months or years to resolve, as well).
But for your sense of size, 500 to 2,000 in unoptimized locales, is your answer.
"You imply ~75 is a better target.
I'm using 75 because that is the bogie that T has set in their Salt Lake City trials, and it is a stated objective for future re-engineered locations, if the trials are successful. The trials I'm referring to also address outside plant electrical powering alternatives, as well.
"Is upstream really that important for say the next 3-5 years? I'll bet 95% of web users consume almost *no* upstream bandwidth. I know I sure don't. If I had a 1200bps upstream I probably wouldnt even notice.
You are limiting your views to your own personal experiences, and experiences of others you know of in your closed circle of affairs, and that can be an extremely dangerous and costly thing to do for a company that is involved with cable modem services.
In the last two years I have worked on several remote access RFPs for brokerage/banking that have specified upwards of 90,000 users who required work at home communications capabilities. Dialups were no good, except for the lower order echelons (who need it more, but don't get it), while 2B+D ISDN (sometimes multiple bonded 2B+D ISDNs) were specified for the middle management levels and higher ups.
[[I know of one group in the SoHo section of Manhattan in the heart of Silicon Alley whi live in makeshift, newly converted, lofts that are Internet Ready.These jokers have access to their own at home T3 pipes. No fooling around. They attach via coax to a colocation mux in the comms room in the same building, and pay through the nose each month for the privilege.]]
Returning to the national RFPs, Cable Modem was expressly excluded from the bidding field, purely on the basis of this very issue having to do with return path limitations. Anyway, the cable operators would not entertain such a capability using their normal resources, in any event. They are not built for it at this time. But at the same time, look at the level of lost opportunities they are forced to live with.
If I am one practicing consultant who has been exposed to requirements for over 90,000 VPN RAS users in a single 18-month period, then could you begin to extrapolate what the field must be like across several tens of thousands of consultants' experiences where VPN end users is concerned? It's huge, being far larger than ATHM's entire current user base, probably a full ten- to fifteen-fold larger, if I had to guess at this time. How does the number 5,000,000 sound? Can someone corroborate or correct this, please? I've lost track of the growth curve in this respect, with time.
Aside from these national RFPs, I've done what I would consider an extraordinary amount of work from NY with partners on the West Coast and overseas. My email at the present time must carry the load, through attachments and other kludgy methods, but I see a need for collaborative groupware apps, soon, faciitating group viewing of large graphic files, and multimedia is going to be an imperative to replace video conferencing, if not only audio conferencing, which itself is a must. I am not alone in this need, I could assure you. And the trend is nothing but upwards, going forward.
"Also with cable penetration rates less than 10% a 75 home node is really only 8 users maybe a dozen max. Or did you mean 75 active accounts?"
Perhaps the cable operators, T, specifically, are not as precise as I am, if I might borrow one of your observations. Seventy-five is the number being bandied about right now for experimental purposes, and has been earmarked as a theoretical model for consideration for a group of online users.
Ideally, the number of attached homes per any segment or node should be dynamically configurable from the NOC, upon invocation of service. But I don't think they are that sophisticated yet. Instead, it may be a process involving Kentucky Windage and wing-and-a-prayer provisioning, using manual practices at the pole or pedestal level, for some time to come. In other words, it could be that until each area gets settled in, it will be a matter of making manual adjustments on the fly. Good question, though.
In another message [which I deemed benign enough to be considered nonviolative of trust if I presented it here], you asked,
"I've always assumed that the "fiber" part of the Fiber-Coax was digital. In other words the coax ran to a place where there was another modem whose output was shipped onto a fiber IP based network.
This is a common perception, but the fact is that the signal is analog through and through. At least that is how it is on today's HFC architectures.
The information may be digital, as on the cable modem part, but even here it is encoded to digital and transported through the use of by some analog QAM or QPSK scheme, or the like, and sent in RF form back to the head end via fiber from the field node. On the fiber itself, various AM and FM modulation schemes may be used, as well as some vestigial sideband technologies.
These are all duly registered analog technologies, carrying both analog (program TV in NTSC format) and digital (Modem) information over the same RF links at different frequencies (frequency division multiplexing, or [O]FDM).
"... what is happening is that the coax signal is being modulated to fiber wavelengths at local nodes and being sent to the head end in analog over the fiber."
So far so good.
"I assume the head end would then demodulate back to TV frequencies and give it to an electronic based modem (which of course demodulates again). *Then* it is given to a digital fiber network."
Good up until the last part. When the data signals are demodulated at the head end, they are inserted into a local router and forwarded to the core, or over ATHM's backbone. If this backbone is what you were referring to as the fiber network, then you were correct.
Comments welcome.
Regards, Frank Coluccio |
