v4 vs. v6 performance

Dave Wilson dave.wilson at heanet.ie
Wed Jun 7 18:08:15 CEST 2006

Hi all,

In some ways changing the subject - this isn't specific to any particular site,
but a good example was posted to the list, which I'd like to use.

> I've much better connectivity to many IPv6 sites than with IPv4. Example
> IETF 151 ms with IPv6, 124 with IPv6.

I believe (and am guided by this principle in my own work) that our goal now is
to make IPv4 and IPv6 performance identical. If one is better than the other -
it doesn't matter which - then it doesn't indicate that one network is "better"
than the other. It indicates that the topologies are divergent, and therefore
extra work is required to maintain one protocol over the other.

I'm going to take a leap here and assert that a lot of the people who will ever
deploy an IPv6-in-IPv4 tunnel have already done so. To get anyone outside that
category, IPv6 needs to be a service they get alongside IPv4, without requiring
them to effectively maintain two network topologies.

So our emphasis now needs to be on encouraging production deployment in service
providers and systems alongside IPv4, even if that means making your IPv6
performance "only" as good as your IPv4 performance.

> This is always relative. Of course, it depends on how both ends are
> connected, and this will improve with the time.

This will improve with activity, which is not a direct function of time, but
requires us all to make some effort in our deployment and (perhaps more
importantly) in our procurement.

All the best,

dave.wilson at heanet.ie  ------------------------------  PGP key:  1024D/C757ADA9
tel:  +353-1-660-9040  H.323 Video GDS: 0035301101738  5170 1A48 5E62 AD51 FDFF
fax:  +353-1-660-3666  Multiconf GDS: 003530110050738  1A6D 3454 2FA0 C757 ADA9

More information about the ipv6-ops mailing list