Bufferbloat.net’s core services are fully IPv6 enabled.
Do you have Ipv6 to your home or business/university? Native? Or can you get 6to4 to work? 6rd? Or a tunnelbroker.net tunnel? Or?
For those ssh users that lack easy access to IPv6, I plan a “jump” box that is dual-stack, and it’s only the core bufferbloat servers that need to be dual stacked - obviously most of the edge boxes will be ipv4. IF there are websites that need to run on a virtual, I plan to run them through a proxy.
The main folk I’ve been dealing with (jg, linville) both have native IPv6 support, and I’ve been using 6to4 tunneling with reasonable success. In fact, the voip calls we’ve been making over linphone directly over IPv6 have been the best quality voip calls I’ve ever experienced.
Based on comcast’s example of running their entire monitoring stack on IPv6, and the relative lack of easy access to ipv4 for our virtual servers, I have been making a big push to IPv6 enable everything at least on the monitoring side, rather than use private (RFC1918) addressing.
The prospect of coping with private addressing across the planet for virtual servers strikes me as far more difficult than coaxing IPv6 to work well.
So far I haven’t found any service majorly broken, at least on Linux. Tested thus far are apache, fcgi, snmpd, the web interfaces to cacti, mrtg, redmine, ssh (of course) and multiple other services. In fact, I haven’t found anything that broke thus far besides:
??shipka.bufferbloat.net \$: ping6 ipv6.gatech.edu
64 bytes from 2610:148:fd8f:d7fc:203:baff:fe8f:29d: icmp_seq=3 ttl=58 time=64.2 ms??
Very few traffic shapers “do the right thing” when it comes to IPv6 traffic. It is likely that your interactive traffic usually shaped by a traffic shaper like wondershaper, won’t be.
To my knowledge, none of these Linux traffic shapers “do the right thing” when confronted with IPv6 or IPv6 encapsulated traffic.
put the following as your /etc/gai.conf
label ::1/128 0
label ::/0 1
#label 2002::/16 2
label ::/96 3
label ::ffff:0:0/96 4
label fec0::/10 5
label fc00::/7 6
label 2001:0::/32 7
I don’t know why this works, nor do I think it’s entirely correct.