Tuesday, April 24, 2007

 

PON, the Architecture of Control

Somebody recently declared that in ancient Rome, bridge (pont) building was the highest form of technology, and the Pope wanted the bridge builders' glory to rub off on him, so he started calling himself "The Pontiff." I have no idea if this is true.

Martin Geddes wrote to me to point out a very good posting in the Fiberevolution blog that claims PON (Passive Optical Network) is 8% cheaper than Point-to-Point (aka Direct Fiber, aka what we in Baseball Land call Home Run). He asks whether Verizon's shareholders are upset that FIOS uses PON.

(Answer: No. Verizon's shareholders are upset that FIOS has such a slow payback. They should be delighted by PON, but they're not. They don't care that Verizon needs to go to the moon, instead they're happy with a higher tree.)

PON, by the way, is the architecture of control. Fiberevolution says that PON is ideally suited for broadcast, that it "mutualizes" the bandwidth, and that it is particularly unsuited for e.g., when some users use a LOT more bandwidth than others, because it makes congestion for the other users. In contrast, in Home Run architecture, you simply re-light strands with heavy use to run faster.

Actually, this is a strength in the view of the telcos and cablecos. As in, "Those nasty P2P users. We need to detect certain applications and discriminate. We need to prohibit certain kinds of traffic, oh did you want video entertainment? We have that over here." In other words, PON gives the facility operator good reasons to rejoin the content to the conduit, which reifies the old biz model that the "any app over any net" Internet took apart, which they love.

Another "strength." Suppose the tide changes at the FCC to make unbundling (i.e., sharing or wholesaling infrastructure to would-be competitors) an acceptable topic. Home Run is built for unbundling. PON isn't. Facility operators hate unbundling. If they build PON, they can reasonably say, "Unbundling is technically infeasible."

In contrast, in Home Run networks each fiber stands alone. I can have a fiber run by Company A lit at J Mbit/s and my neighbor can have another fiber run by Company B, lit at K Mbit/s. Home Run (or Point-to-Point, or Direct Fiber) provides a smooth upgrade path, serves users with different needs, and grows as our need for throughput changes. Home Run is the way to maximize the network's benefit if one expects the future will be uncertain or rapidly changing. But the facility builder (tellingly, "owner" and "operator" are not good words to use!) loses control of the asset.

So having the control PON architecture provides is VERY valuable to facility operators because it provides the unfair advantage to which telcos and cablecos have become accustomed. Even if PON were 8% **more**expensive** it might be the telco-cableco architecture of choice.

Technorati Tags: , , , , , ,


Comments:
David,

Thanks for the comments. You raise a good point, and I may very well be biased towards open service models that - as you rightly point out - incumbents and large players tends to dislike. I'll be writing more about business models in the coming weeks.
 
Two comments:
1) There are ways to lay out a PON network which still allow for unbundling. See http://web.si.umich.edu/tprc/papers/2003/241/Banerjee_Sirbu.pdf

2)The technological upgrade path for today's GPON is wavelength per customer (currently being trialed in Korea) which eliminates contention downstream of the central office and allows for different bit rates per customer on the different wavelengths
 
Why bother to investigate true life cycle costs when attributing sinister motives is so much easier? PON is preferred since it can show lower longer term costs than p-to-p. Just ONE example: Repair. When (not if, WHEN) a cable is cut PON recovers 32 or even 64 services with 1 single mode splice, in PtoP it's 1 splice recovers 1 service.
 
Interesting points about PON.
In France, incumbent telco France Télécom has clearly expressed itself in favor of the PON, whereas one of his two main competitors, Free (Iliad) has chosen P2P. The 3rd big one, Neuf Cegetel, says will use both architectures, depending on the situations.
 
Post a Comment

This page is powered by Blogger. Isn't yours?