Hi, On Mon, Apr 23, 2012 at 01:33:00PM -0700, Dave Taht wrote: > > (And no, I wouldn't advocate 6to4 being enabled by default anyway). > > I had it enabled by default during comcast's trials. It worked great, > on their network (and having a /48 was good too). It didn't work very > well elsewhere, so it's in there, but disabled by default. > > I also have HE tunnels and 6rd working. > > I'm more of the opinion that ipv6 needs to be made to work, using > every method possible, and if possible, those methods should be able > to co-exist, using policy routing. That proves to be hard.
We're getting a bit off-topic, I'm afraid, but I need to add a rant
here :-)
I think that HE+6rd are useful transition technologies, while 6to4 is
harmful, and "no IPv6" is much better than "only via 6to4".
Let me explain:
The problem with 6to4 as the sole connection "to the IPv6 world" is the
anycast relays, so you have no control over the return paths - *and*, to
add insult to injury, people are known to drop proto-41 packets coming
*back* from the relays, so users end up with blackholes. So "SYN -
timeout - SYN - timeout - SYN - timeout" instead of "SYN - immediate
'no route to host' - fallback to IPv4" if you have no IPv6 at all.
There are no IPv6-only services out there today, and turning on 6to4 with
its associated risk of having unreliable and slow connections to random
*parts* of the IPv6 world is doing IPv6 a disservice - users will discover
"things get faster if I turn off IPv6", spread that lore, and that's not
useful.
(Now there are certainly cases where 6to4 is useful - if you are talking
6to4-to-6to4 only, routing only 2002::/16 into the 6to4 tunnel. But please
do not, never ever, point a default route to the 6to4 tunnel, unless the
user has explicitly asked for it).
(*) 6rd is OK, even if using the same underlying "tech" as 6to4, because
unlike 6to4, it knows it's relay, and the path to and from the relay is
inside the ISP's domain - so you actually have someone who can troubleshoot
issues, and latency is likely to be as good as IPv4.
gert
--
USENET is *not* the non-clickable part of WWW!
//www.muc.de/~gert/
Gert Doering - Munich, Germany g...@greenie.muc.de
fax: +49-89-35655025 g...@net.informatik.tu-muenchen.de
pgp9CNxWxuUdp.pgp
Description: PGP signature
_______________________________________________ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/mailman/listinfo/openwrt-devel
