Although the contents of IPv4 and IPv6 routing toolboxes are practically identical, we can still observe significant differences in the behaviour of the IPv6 routing system, compared to its IPv4 counterpart.

First of all, IPv6 Internet is still by several orders smaller. According to the CIDR report (data from February 2005), inter-domain routing tables contain more than 150 thousand IPv4 network prefixes but only about 700 IPv6 prefixes. The complexity of IPv6 routing is thus incomparable to its IPv4 counterpart.

IPv6 routing still suffers from two characteristic problems:

1. Network interfaces with multiple global IPv6 addresses are prone to asymmetric routing,which is often undesirable. This is mostly due to improper implementation of source address selection.

2. The IPv6 Internet is still, by and large, an overlay network over IPv4 and its tunnels distort the inter-area link metrics (number of hops in AS paths) and mislead routing algorithms.

Both problems can be expected to disappear after IPv6 becomes more mainstream and the implementations stabilise. One highly visible difference is likely to persist though: most routing domains in the default-free zone filter out all incoming prefixes that are longer than /32. Such a strict aggregation of prefixes is absolutely crucial for long term viability of global IPv6 routing. Note that the set of /32 IPv6 prefixes still has the same size as the set of all IPv4 addresses, so the aggregation rules may perhaps become even stricter in the future.