For those of us who remember such things as a 1200-baud data connection over a hissy, dial-up phone line, this may be hard to digest, but here it is: The “high-speed” part of “high-speed data” hardly matters anymore.
What matters is capacity and throughput. Fast is fast, and fast is nearly ubiquitous in North America. When was the last time you complained about a slow connection (hotel rooms excluded)? When was the last time a reboot of the modem didn’t fix it?
The marketing of data services, from the days of the dial-up telephone modem to now, uses speed as the basis for greatness. 1200 baud, 2400 baud, 9600 baud, all the way up to today’s offerings of 100 Gigabits per second and higher.
But there comes a time when the speed gains just aren’t noticeable anymore. At some point — let’s say 50 Mbps, like several operators now offer as a high-end tier — it’s difficult to discern whether that Web page really loaded any faster.
Consider a home with five HDTVs, all on, and a cable modem attached to a wireless router spraying signal to five IP devices, all doing something big — streaming video or backing up files to a cloud-based server. In the same home, five VoIP phones are all in use.
All in, that house is consuming perhaps 30 Mbps of capacity, in that moment (assuming MPEG-2 compression on the video). Yet the home’s subscription tier supports, say, 10 Mbps downstream and 2 Mbps upstream.
Herein lies the difference between “speed” (10 Mbps) and “capacity” (30 Mbps of usage on all screens). One measures how fast one machine connects to another; the other measures how much stuff one can push through a connection.
Going forward, “fast” will be assumed. The differentiator will be the ability to serve up speed to the increasing number of things in our lives that require or work better with an IP connection.
Already, MSOs are anticipating an average of six IP-connected screens in homes by 2015; some friends in this geeko- sphere already count 70-plus.
For cable, discussions about network capacity for IP-delivered services correspond to the channel-bonding feature in DOCSIS 3.0. Already, some operators are bonding as many as four digital channels to carry IP-based traffic — Web browsing and voice, of course, but also subscription video, both linear and on-demand.
Ultimately, to simulcast the linear lineup in IP, they’ll need to bond 24 to 30 digital channels
By now, most major cable providers are mostly deployed with DOCSIS 3.0 capabilities. That’s good, because it’s designed to withstand what’s coming. In speed, and in throughput.