When I first started this website, the so called ‘exaflood’ was looming on the horizon with all sorts of dire but ultimately silly predictions. According to Cisco, in a few short years the ‘exaflood’ era will give way to what we thankfully aren’t calling the ‘zettaflood’, at least yet.
In its most recent Visual Network Index forecast, Cisco says that in 2018 (just four short years) the internet will be handling something like 1.6 zetabytes per month. That’s a lotta bytes, but it’s not quite a yottabyte – which will be the next numerical prefix at which we can all bow before the power of exponential growth.
And lo and behold, the internet hasn’t broken yet. Well, unless you count when peering providers *choose* not to upgrade their interconnections to pressure each other into paying etc. Technologically and financially, networks are having no problem keeping pace with the upgrades needed to carry all these bits.
Actually, it’s interesting that for all the traffic growth we’ve seen in the past six years the gear has only jumped in raw bandwidth per wavelength from 10G to 100G and that transition is really only just getting started. Oh they’re working on the next wave, whether it’s 200G, 400G, 1Tb, or whatever.
We certainly needed the bigger pipes in the core, but it has to be said that we didn’t need them as badly as we thought we would. Much of the growth has been at the edge, with CDNs and other means of localizing content absorbing the brunt of the growth.
While we’ve got a lot more bits being transferred, I wonder if anybody has done any studies on how the average distance traveled per bit has evolved over the years. I suspect it’s down quite sharply, which is at least as important to understanding the evolution of our infrastructure as the yearly tally and projection of raw bits per month Cisco puts out.