On Fri, 2010-02-12 at 12:13 +0100, Marco Hogewoning wrote:
I guess the youtube traffic is there, it must be there, but I haven't seen any. Even on my own link at home it won't show up because those little peaks on the graphs get blown away by other traffic and since I'm the only user I know which is which :)
If you want measurements, measure at the end points that is the only reliable method. Make sure everybody measures the same, using the same set of counters (either byte counters or flow sampling) or find a way to correct your data between the 2.
Well, as you say, we really don't have to overly complicate this wrt. Youtube IP traffic. I doubt anybody will argue that eyeballs have traffic to Google/Youtube traffic. How much? I remember some flow sample statistics from a few years back here from Gothenburg, covering some ~10k students with plenty of bandwidth. I assume these _swedish students_ knew how to use P2P. :) Google+Youtube was about 3rd or 4th unique origin AS by bytes transferred! I don't have the references available, but this I personally consider a ~fact for eyeball networks, at least in Scandinavia. Prove me wrong. :) I don't remember what this meant in terms of fraction of the total traffic, but I suspect somewhere around 5-10%? If what we want to find out is % IPv6 traffic out of all IP traffic, and how this might change by some large players, or protocols, (suddenly) changing behaviour, this is relevant. I assume the ranking of Google/Youtube has dropped much since then. More data needed. I could write more, but where I want to come basically boils down to that I strongly suspect that the current IPv6 support infrastructure for the transition mechanisms today do not support a Google/Youtube ceasing to use the DNS resolver whitelist method as a connectivity quality assurance technique. And I believe we will have to cope with the transition mechanisms for quite a while. We're a long way from native IPv6 to a majority of Internet users. But OS:s support transition and this is being "rolled out" every day, everywhere. So, the operator community should IMO continue to do more to help the transition. What would happen if $LARGE_FRACTION of p2p traffic suddenly moved to v6 transitions? This can happen. So it is, like you say Marco, not very significant how many % traffic there is today. There are much more important aspects to look at. I fully agree with you on this. Basically what to follow closely, I think boils down to something like: * users type of v6 connectivity, if at all * changes to p2p * changes to ~top-10 CDN:ish networks (youtube included) * rest of the web in general There are plenty previous work done in following/measuring several of these points. Would be nice with some potaroo.net type, stable, place to follow it, though. Regards, -- Martin Millnert <millnert@csbnet.se>