IPv6 Ignorance

Jeroen Massar jeroen at unfix.org
Tue Sep 18 11:52:54 CEST 2012


On 2012-09-18 11:39 , Ole Trøan wrote:
[..]
> there is nothing intrinsic in a 1280 MTU or a dynamic prefix with 6rd.
> that's if so a deployment choice. you can very well get a dynamic prefix with "native IPv6"
> in the access network too.

I am very aware, but it is not how they get configured, even though they
could. That is, I have not seen 6rd being deployed differently, even though

(and typically it is so they can sell a business variant)

> 6rd offers native IPv6 on the LAN side.

Any connectivity that is not encapsulated any more does that ;)

> as an end user, why would you care how the traffic was carried in the network?

I don't.

I do care about the MTU though, it does make a tiny difference, and that
I do not have to update my prefix for all the hosts I have at various
locations. (which could be scripted, but I do not like having to
auto-update my now static-served DNS; which might have to change once
due to DNSSEC though...)

For the German speakers/hearers, around 03:19 they mention why it is so
useful to have static addresses: you can just have a roadwarrior and
always connect back home or everywhere as the addresses are static, thus
that filters much easier...

http://cre.fm/cre197#t=02:56:38.698,03:23:11.323

(long live AYIYA for that one ;)

> is PPPoE over L2TP "native"? is anything over an MPLS tunnel native? ;-)

Well technically DSL is also tunneled inside ATM inside voiceblabber ;)


On 2012-09-18 11:42 , Nick Hilliard wrote:> On 18/09/2012 10:39, Ole
Trøan wrote:
>> as an end user, why would you care how the traffic was carried in the
network?
>> is PPPoE over L2TP "native"? is anything over an MPLS tunnel native? ;-)
>
> You care when your tunnels cross third party infrastructure.  There's
> no problem running services over 6rd or mpls or other interior
> tunnelling mechanisms.


Please note that I have been running tunneled IPv6 over 3rd-party
infrastructure for more than 10 years without any issues.... yes it
might break, but over those 10 years it might have been broken 1 day in
total and a huge learning resource.

Yes a tunnel is not always optimal, but it is better than the current
alternative: a tunnel with lower MTU and dynamic prefix.

That one day will hopefully change though and likely I'll then simply go
for a business account or so that I do not have the changing prefix issue.

Greets,
 Jeroen



More information about the ipv6-ops mailing list