I am not arguing I am just curious why teredo is more efficient than most tunnel brokers. I thought 6to4 and/or 6in4 had a lower protocol overhead?
Efficiency aside for any site that I have tested HE's peering agreements seem to produce lower latencies and better throughput than the IPv4 network paths provided by TWC. But this might be unique to my situation due to how close I am to the HE pop in chicago.
As far as path length is concerned it is my experience that HE.net has better peering agreements and faster interconnections between routers than TWC. I live in Syracuse, NY and regardless of destination all of my traffic is routed over TWC's network until it reaches Chicago. Upon reaching TWC's infrastructure in Chicago my traffic can be handed off to Cogent, Level3, Telia, etc or continue on TW's network. Or as luck would have it can be handed off to the HE POP in the same data center as the first hop in Chicago. Traffic that goes through the HE node has lower latencies and higher overall throughput. This is not just a general impression or anecdotal wget -4/-6 benchmarks. I have a local stratum one ntp server (GPS+PPS) for stability/robustness (and because I was interested in the overhead of the tunnel) the machine is configured to query a number of dual homed S1 ntp servers. The ipv6 associations consistently have lower delay and jitter statistics than their ipv4 counterparts.
I am not a teredo expert but I thought teredo used relays for some portion of traffic?
The second bit has nothing to do with efficiency and its worth pointing out that HE's tunnel broker service is not motivated by altruism.
Efficiency aside for any site that I have tested HE's peering agreements seem to produce lower latencies and better throughput than the IPv4 network paths provided by TWC. But this might be unique to my situation due to how close I am to the HE pop in chicago.