E.Davies, this replies to both the upstream link you sent and to "some" of the questions in you PM along with those of others, concerning pure.. and certain ATHM issues.
>> Has optical reached the point where it can "understand" the data embedded in the optical? How can you know where & how to route the data without converting the data to electronic form to translate the headers, look it up in a routing table etc? How can you store & buffer optical?<<
Optical elements will have logical capabilities embedded in them, most likely through purely electronic means at first, and later as optics improve, these will be displaced increasingly by hybrid and then purely optical techniques in some situations.
I don't see the fundamental protocol (the rules of traffic as defined by the IETF in the IP realm) changing radically at this time. I do see alternative means, however, of representing the line code at the physical layer..
For example, one router manufacturer is talking about using a protocol similar to Cisco's Tag Switching, or its cousin Multi Protocol Label Switching or MPLS. Instead of staying with the status quo line coding of IP in the T-1 and T-3 space, using electrical or optical bi-polar representations of ones and zeros in the data and header information, they will optically "bar code" their wavelengths through spectral manipulation, in a more organic fashion, in a kind of "multi-protocol label switching" MPLS scheme. In this manner, entire "flows" will be directed (routed) to designated end points or nodes. It's not perfected yet, far from it, but this should give you an idea of where some folks are headed on the road to optical purity.
>>I'm still looking for a reply to my longer post especially about what ATHM does & does not own and what exactly they use their network for.<<
Some of my opinions and ultimate confusion that I share with you on this can be found in my post # 4023. Others have been answered here, kinda, in the replies to that post.
My take is that ATHM finds itself in much the same circumstances as SIAC, the Securities Industry Automation Corporation, the quasi-autonomous subsidiary of both the NYSE and the ASE that supplies ticker quotes and performs daily clearing and settlements.
If you drill down into each of the latter NYSE and ASE orgs, you find that even these institutions are, in kind, made up of member brokerages and securities firms, themselves. Absolute ownership becomes a moot point at some level, as the subsidiary has more important things to think about from day to day as trading volumes go through the roof (do you see the analogies here with bandwidth demand?). They serve everyone qualified to be on their rosters, in effect.
>>4) How will ATHM develop past the current contract restrictions to what the internet needs to be?<<
Several here have stated that ATHM will leverage all of the powers of each of their owners and partners. That sounds great, phenomenal, in principle, but management through consortia has its pitfalls and associated delays, I've found. The "One roof" concept is starting to have a better ring to it, the more I look at it.
>>The MSO's owns the wires. Who owns the electronics of the distribution points? Who owns the architecture of the local network?<<
To the best of my knowledge, the MSOs do.
>>How much of the networking is done at that point or are these points basically "dumb"?<<
Basically passive, dumb elements in the field, with increasing intelligence introduced at the head end (hubs and routers, possibly some servers) and the ultimate level of directory and content services at ATHM's POPs [and beyond, when linking to outside >off-net< services is required].
>>Where does TCI stop and ATHM start?<<
Don't know exactly, but the field cabling and the head end is TCI's, and the POP provisions belong to ATHM. The T1/T3/OCx links between them are unclear to me at this time. I doubt seriously if a general rule even applies at this stage, because of the number of participating partners and their unique circumstances.
>> I want to know most of all because when service becomes overloaded (and it will) who is responsible for modifying the local distribution to balance/divide the local nets to increase performance? <<
The turnpike effect on the greater net will never be entirely solved, only mitigated in stages. On the local nets, there are industry-wide initiatives, which are still in their formative stages of mobilization, taking place right now to address the segmentation issues. Segmentation amounts to breaking up the single collision domain with many users in the serving area to a number of independent segments with fewer users, in order to alleviate future bottleneck issues such as those you've described.
I'm not certain, but I think that these initiatives are under the direction of CableLabs, as we speak. Just how expensive this will be remains to be seen, and I can't comment on just who it will be who will follow these recommendations. It will likely take place on an individual case basis at first.
>>It also has a lot to do with how easy it will be for someone else (like say AOL) to jump into the same space and compete.<<
The unbundling issue again. I don't know. IMO, a lot of this is going to depend on how successful T is in demonstrating "structural separations" between their (T/TCOMA's) voice and other services that have traditionally fallen under the rubric of common carriage, and those which are relatively immune from regulatory interference under the heading of information services.
>>I think you implied that you think the ATHM national network isn't used for bypassing the internet backbone and it's limitations.<<
I didn't say that, exactly. In any event, I have since retracted that "notion" that some have picked up on (my point had to do with the incremental enhancements in recent press releases, and not the status quo), and clarified my acknowledgement that ATHM is indeed thinking in the proper terms when it comes to circumventing the major Internet NAPs.
>>Do you think my picture is wrong? <<
Admittedly, I don't have inside information at this time. So, I can't intelligently comment on what you've stated re their networking strategy. But most of it seems to be in line with what I'm learning here.
>>I'm sure 2.5Gb links is only a start, how high do you think they need go and how achievable is it? I dont have enough feel for a network to guess how many users would use one link at any one time to estimate. <<
The answer to this, again, is in large part dependent on what services ATHM is permitted to deploy on its own, in addition to which services the MSOs and T will deploy. Bandwidth allocations may become an issue at some point, if they're going to go separate ways.
How much of ATHM's services will remain residential in nature and consumer oriented as opposed to the @Work or other future components... all of these could potentially enter into the calculus. They probably have, already, on the planning side, if not the design side.
In a general sense, however, I'd agree that the 2.5 Gbps backbone is indeed only a starting point. A sense of balance must be reached, however, and in so doing may mitigate some of the demands on the backbone if the end points continue to be served by collision domain technologies. The alternative would be to move closer to a deterministic, or, better yet, purely optical, space. Neither of these seems to be on the radar screen at this time, however, from what I've read or discussed anywhere. They will more likely optimize their last mile techs in accordance with measures I've already outlined above: re-segmentation.
>>So far it's not been clear to me to what degree ATHM intends to be more an ISP (like say Mindspring) vs. a content provider (like AOL)<<
All of the above, I think, and then some.
----
The remainder of your post centers on the potential conflicts of interest that may ensue in the delivery of services vs the proprietary interests of the owners. I don't think that my speculating on this aspect any further beyond that which I have will serve any benefit. I'm in the dark on many of the future directions in that regard.
Hope I've answered most of your questions adequately.
Regards, Frank Coluccio |
