Measuring the performance of world IPv6 day

Despite that IPv6 day was held in a dual-stack configuration (as only parts of the Internet went IPv6 enabled), the metrics obtained provide an insight into what we might expect when the day comes to transition fully.

Information published by the RIPE Network Coordination Centre (RIPE NCC) show that in the dual-stack configuration, there was a measured slowdown on the IPv6 side of the equation, however the difference in slowdown isn’t too major.

      [ Source: RIPE ]

So what happens when a dual-stack client is asked to resolve an address with both an A (v4) and AAAA (v6) address?

If a dual-stack client connects to a hostname for which both an A and a AAAA record is advertised in DNS, it has to make a choice: Connect over IPv4 or over IPv6.

Well, as you will find in this post by RIPE, there are a number of factors involved in determining the course of action.  These considerations, broadly speaking are forwarding (e.g. by routers, or by hop count), end points (where different protocol endpoints may resolve to completely different hosts) and network topography (differential treatment of packets, etc).

The measurements performed on world IPv6 day measured ICMP/ICMPv6 packets and not broader protocol traffic such as FTP, HTTP etc.  There was also a significant difference in network endpoints for IPv6 enabled sites (which wouldn’t necessarily reflect an all-IPv6 network). 

[ Source: RIPE ]

Adjusting the measurement (above) to only include IPv4/IPv6 sites which were not separated (where the IPv4 and IPv6 end-point are at the same physical location) the difference drops a little bit.   Anyway, it’s an interesting set of data, and time will tell what bodes for the future..

More Reading

Leave a comment

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.