Wednesday, April 4, 2007

Unfortunately, FTTX Doesn't Tell You Much

At some point, there will be a pretty compelling business difference between a fiber to neighborhood, fiber to a node of 16 to 32 homes and fiber directly to the customer. Standard definition TV may persist for awhile in closed environments. Cable operators might even see some business advantage there, as customers can continue to use their analog TVs without any new decoders to receive at least some programming. Of course, the counterweight is that bandwidth taken up by analog prevents efficient use of potential digital bandwidth to provide new services.

But when all terrestrial broadcasting instantly converts to HDTV in a couple years, and cable programmers match the move to avoid being seen as inferior in quality, most linear programming will be in HDTV format (the issue is what cablers will want to do about "standard definition" NTSC signals).

The assumption that just one concurrent HDTV stream will be required by any single home has to be suspect. To some extent, the access network has to be built to the level required by the most demanding user. And while it still makes sense to retain the ability to "spot upgrade" on an incremental capital basis as more demanding customers are signed up, the access network has to support such incremental and spot upgrades. Whether or not all the theoretical bandwidth is presently required, the ability to provide it on demand is something the access network must be built to support.

The option to deploy a "thin fiber" network saves capital investment for the moment, to be sure. Planned properly, fiber then can be incrementally extended deeper into the network. Such rework remains a non-trivial exercise, though.

Also, keep in mind that all discussions about what video might do the backbone of a network is a separate matter from what might be required in the access network. The reason is satellite delivery, regional server farms and other ways of substituting storage for bandwidth, processing for bandwidth, non-real-time delivery (store and forward) and alternate networks. All of these forms of substitution allow for easing the stress on backbone networks. The amount of stress on access networks typically is the real pinch point.

No comments:

Computing Archiitectures Now are Dependent on WAN Performance, Not LAN

These days, computing performance mostly hinges on the wide area network, not the "local" area network, a big change from earlie...