Starting back to front; that's a good question. If we could gain insight into APNIC's chosen ad syndicator, we'd know; APNIC is the RIR for the East, and these stats were initially designed to help them track performance in their areas; not on a global scale. This can be seen by the fact that APNIC's results are almost double the Akamai stats for WA and Afrihost who just made it onto Akamai's board (I can't comment on our performance on Akamai because it's based on unique samples). I'd trust Akamai more as they serve one of the largest CDNs in SA - DSTV Now (which still doesn't support IPv6 ironically).
The numbers are relatively low on APNIC's list, specifically for WA. Vuma reach has a massive effect in terms of sample- this is how you've seen two meteoric rises in v6 across both APNIC and Akamai's results. IPv6 is enabled by default on Reach (so all clients on supported ISPs (WA and Afrihost) get v6), to tens of thousands of eyeballs. We're not on the reach network (very low margin and massive volumes means Vumatel isn't considering smaller ISPs for this network at all). But tens of thousands of reach homes, shared between a handful of ISPs, mean subscriber numbers are skewed relatively higher than ISP distribution on other FNOs. Population density around reach also means more active devices per home. Assuming 5-8 devices per home (which is reasonable for your average 4 person home), all running v6; the sample size collected by APNIC is still relatively small considering the reach of reach (no pun intended). WA for instance, at 23k samples, and assuming 7-8 samples per home (7-8 devices), means only 3-3.5k homes - and their reach footprint is significantly larger than that; and we're not even looking at their L2 and trenched homes. I'm just saying, the stats are useful, but certainly not an absolute reflection of IPv6 usage, adoption or ISP size for any network in question. I'm not going to bash APNIC; considering it's one of two unique sources of information on this matter; I'd rather see some stats than none at all - and admire their effort. Some transparency regarding data collection methodology would be cool. I don't think we're altogether worried about it being a representation of our network stats.
Considering the number of eyeballs on our network, IP utilisation stats, no, I don't think the scale comparison is precise. I personally have more than 13 active IPv6 devices on my LAN at home, so do you. Our office currently has over 50 unique IPv6 sessions active. I can go on, but you get the gist - something's not adding up. Maybe we all just don't like a specific cat video? Without insight from APNIC, I can't tell you more. I'm not fretting about APNIC's results; we're seeing every new L2 client run v6 just fine, we're seeing a growing amount of v6 traffic from older L2 clients as well as a surprising amount from FTT
), but it has been growing steadily on our side and we are happy with the progress. We don't remotely control most clients' routers, this means relying on clients to enable IPv6 on their routers (which has been slower than hoped). We have enabled IPv6 on routers we do manage and do enable IPv6 by default on new clients, so we expect to see more eyeballs on APNIC's ads soon enough.
Sign up with Google