Thursday, October 16, 2008

Exaflood, schmexaflood?

I just got home, having been out at a couple of interesting events today (which I will blog after the kids are in bed). However, I notice that the awesome Kenjiro Cho and the IIJ team have once again produced what looks to be a very interesting paper on residential bandwidth consumption in fiber-rich Japan. I haven't had a chance to read it, but the abstract seems to suggest that their results may moderate the exaflood argument somewhat. 

Also waiting in my inbox this evening is an awesome-looking set of video presentations from ARCEP's seminar on ultrabroadband - there goes my evening!


Andrew Schmitt said...

Thanks for the link to the updated report.

I looked at this in another fashion and concluded that once subscriber growth was backed out, per subscriber usage was increasing at 18% annually in Japan.

Dean Bubley said...


Interesting. I had a quick scan, but the one term I didn't see was IPTV.

It seems to be covering just the Internet-destined traffic on the broadband line, measured at the edge router, rather than anything else broken out at the DSLAM / head end.

Also, I'm not sure if there is unbundled local loop (or fibre+wholesale) in Japan, which might decouple subscriber numbers from physical fibre/copper deployment.

The other question is whether Content Distribution Networks are included in their peering stats, or whether that traffic is siphoned off elsewhere & doesn't show in the numbers.