ipv6 transit over tunneled connection
owen at delong.com
Fri May 14 22:58:06 UTC 2010
On May 14, 2010, at 1:36 PM, Jared Mauch wrote:
> On May 14, 2010, at 3:43 PM, Brielle Bruns wrote:
>> (Sent from my Blackberry, please avoid the flames as I can't do inline quoting)
>> Native IPv6 is a crapshoot. About the only people in the US that I've seen that are no-bullshit IPv6 native ready is Hurricane Electric. NTT is supposedly as well but I can't speak as to where they have connectivity.
> I can say that we (NTT) have been IPv6 enabled or ready at all customer ports since ~2003. Anyone else who has not gotten there in the intervening years may have problems supporting you for your IPv4 as well :)
>> Being that there's issues that leave us unable to get native connectivity, we have a BGP tunnel thanks to HE (with a 20ms latency from Seattle to Freemont).
> You should be able to get native IPv6 in Seattle from a variety of providers. If you're not finding it, you're not really looking (IMHO).
Depends. If he's in the Westin or some other colo, sure. If not, he may have last-mile expenses that exceed sanity for his situation leading to a tunneled solution.
>> Tunnels suck if not done correctly. We sometimes have faster and more reliable connections through IPv6, so ymmv.
> The tunneled part of the "IPv6" internet fell to the wayside a long time ago, there are stragglers and I have even seen people try to peer over tunnels in 2010, but anyone still adding that level of overlay (v6-over-v4) may find themselves in a world of hurt soon enough.
I have to disagree with you here. Given the proportion of the IPv6 internet that is still connected via tunnels, your statement simply doesn't really hold.
I will readily agree that where possible, native connections beat tunnels. However, tunnels can be a cost effective alternative where native connectivity is not yet readily available and they still work quite well if properly configured and structured.
More information about the NANOG