Re: [MBONED] draft-zhou-mboned-multrans-path-optimization

[Stig Venaas <[e-mail]>]

One thing that perhaps worries me here, is that I think you
may end up translating multiple times between IPv4 and IPv6,
even if no translation is necessary.

I don't think you should simply compare metrics for IPv4 and IPv6 and
choose what is best. Or at least, there should be some additional cost
for the translation that is added to the metric. Doing translation adds
delay (and I would think jitter). Then there is also the issue of load
on the translator.

Stig