[LINK] The PLAN, and broadband speeds?

[Glen Turner]

On Wed, 2007-06-20 at 15:43 +1000, David Lochrin wrote:

>    Let's say the proposal is to replace all existing copper subscriber
> connections with fibre of some sort.  All sorts of potential uses might
> be identified, but the main use would probably be TV (*).

Sorry to disagree, but I've got fiber-to-the-office running at
a gigabit and I never use it to watch TV in the traditional
sense.

If I pull up the traffic graphs the peaks are from video
conferencing.  This suggests that a new customer access
network needs to have enough bandwidth planned to run high
definition video conferencing at phone call quantities.

The other peaks are from big transfers of data and from
YouTube -- the new TV?

Anyway, it's plain that ADSL2+ doesn't cut it as the end-game
for the customer access network. Asymmetric doesn't cut it,
and 24Mbps isn't enough.

I'm not sure if a Fibre to the Node network cuts it either
as the end game. You're looking at about 10-100m of Cat3
lead-in cable. You'd be doing well to get 100Mbps down
that.

If you add it up
 - two HD TVs
 - two HD videoconference
 - some web
 - phones
you need about 1.2Gbps -- call it 1Gbps for convenience and
assuming you can squeeze more performance out of HD video codecs
as CPUs improve. So we're talking fiber to the home as the end game. 
Note that high definition has a heavy use of bandwidth, without that
you get about 60Mbps for the above list. But I've seen HD
videoconferencing -- unlike HD TV the difference is compelling (eg,
you can hold up a page and the other person can read it).

I'm not sure why you'd want to fund an interim customer access
network that can't grow into the end game such as WiMax or ADSL2+.
FTTN makes some sense, in that you can get some revenue in and
convert FTTN to FTTH as demand increases. But you run the real
risk with that strategy of simply paying twice.

-- 
 Glen Turner