The day after World IPv6 test Day, it looks as if the Internet 1.0 has been getting back to its IPv4-borne normality. Monitoring centres report that the IPv6 has one again retreated to the shadows with traffic “falling off a cliff.”
For the 24-hour test period, which ran from midnight UTC (ET+4) on 8 June, IPv6 traffic appears to have roughly doubled, as perfectly illustrated by the spike recorded by network infrastructure company Akamai, reaching perhaps fractions of a percent at peak.
Saying that IPv6 traffic at least ‘doubled’ is not the whole story however and contains good and bad news for those who hope that IPv6 migration can happen sooner rather than later. Indeed, traffic doubled but mainly because it was already at incredibly low levels that barely registered. Privately, many engineers are not even convinced that it is possible to accurately measure IPv6 traffic at all so insignificant is its presence.
On a positive note, measuring traffic against the Internet as a whole is misleading because some parts count more than others. Google’s servers are a part of the Internet that count more than most and the company recorded IPv6 traffic as hitting as much as 0.3 percent, a huge jump for a protocol that until a day ago had barely been used to do more than carry out engineering research.
If 8 June proved IPv6 can survive the real world, the problem is what appears to have happened next. Hopes that content delivery networks (CDNs) would keep the protocol switched on have been dashed, with every recording centre registering IPv6 melting away from the first minutes of 9 June as rapidly as it had advanced only the day before.
Akamai’s graph shows the plummet, most of it in North America where the most enthusiastic big-brand participants – Google, Netflix, Yahoo, Bing and Verizon – seemed to have just as eagerly flipped the switch back to the ‘off’ position, job done, the first minute they could.
“It’s going to be a long slog to IPv6. But for a 24-hour period the IPv6 network looked like a regular network,” says a reflective Rob Malan, co-founder and CTO of Net security company, Arbor Networks.
Until 8 June, IPv6 traffic had been categorised as “other”, says Malan, which was a polite engineering way of saying that it was not doing anything useful that anyone could explain. Throughout 8 June, for the first time in history, it suddenly shifted useful packets, a huge advance it its way.
“For a brief moment the IPv6 network looked like the Internet,” says Malan. “IPv4 has got a long life ahead. With scarcity will come better allocation of resources. We are going to be able to get away with IPv4 for a long time,” he adds as a heavy qualification.
Others remain more upbeat. “Despite the low traffic volumes I believe it was a success,” says Blue Coat Systems’ chief scientist and acknowledged IPv6 expert, Qing Li. “We are talking about IPv6 now. That is the success – it is no longer a backroom discussion.”
“IPv6 was designed not just to solve the addressing issue but to make packet routing much more efficient,” points out Li, who believes that its supremacy is only a matter of time. Until then, people need to prepare to run the two network addressing schemes in parallel.
For many participants, however, World IPv6 Day has come and gone and they have in different ways been left to pick up the scattered packets, musing over what it all meant beyond a welcome burst of publicity.
“I keep thinking of the old wagon trains that left for California. It took forever for them to get to the promised land. It’s going to be a long, hard slog,” says Arbor’s Malan.