IPv4 Is Not Enough

Last week in Chicago, at the annual SIGCOMM flagship research conference on networking, Arbor collaborators presented some exciting developments in the ongoing story of IPv6 roll out.  This joint work (full paper here) between Arbor Networks, the University of Michigan, the International Computer Science Institute, Verisign Labs, and the University of Illinois highlighted how both the pace and nature of IPv6 adoption has made a pretty dramatic shift in just the last couple of years. This study is a thorough, well-researched, effective analysis and discussion of numerous published and previously unpublished measurements focused on the state of IPv6 deployment.

The study examined a decade of data reporting twelve measures drawn from ten global-scale Internet datasets, including several years of Arbor data that represents a third to a half of all interdomain traffic.  This constitutes one of the longest and broadest published measurement of IPv6 adoption to date. Using this long and wide perspective, the University of Michigan, Arbor Networks, and their collaborators found that IPv6 adoption, relative to IPv4, varies by two orders of magnitude (100x!) depending on the measure one looks at and, because of this, care must really be taken when looking at individual measurements of IPv6.  For example, examining only the fraction of IPv6 to IPv4 Continue reading