Re: [v6ops] I-D Action: draft-ietf-v6ops-ula-usage-recommendations-02.txt

[Victor Kuarsingh <[e-mail]>]

Lorenzo,


On Thu, Feb 20, 2014 at 9:58 AM, Lorenzo Colitti <[e-mail]> wrote:

>
>
> I will be the first to support a draft that documents *real* experience of
> ULA in a *real* deployment. But documenting use cases without having
> actually used them for real... sorry, that's hubris.
>
>

I am aware of at least one use real case that I have experience with for
ULA-only.  This use case is for Cable Modem Management.   I guess I should
provide text to the draft on this (as it's a use case for "Connected
Network: ULA-Only deployment").

In this case the IPv6 device as no global connectivity requirements and the
use of the IP address for management purposes is very well scoped.  Using
ULAs in this case is quite safe since the devices will never need global
connectivity (in fact its specifically guarded against).

Using GUA in this case is possible, I know the blowing 1000s of /64s likely
won't get anybody excited, but managing the filters on the modems is much
easier with ULAs.  Using GUAs, one may need to add a few filter lines in
the modem config which is a precious resource (config file space).  Whereas
doing a deny FC00::/7 is easy and very clear.   It's also visually obvious
to the operator which is important in troubleshooting (where you have 100s
of ops people) It's a clean use of the ULA space.

As for documenting use cases, my input is

1) we should document real world use cases where we have real experience (I
am a bit worried about what-if or no-experience areas)

2) I think we should soften the language of the draft to not "recommend"
ULA usage, but more along the lines of if we use them, here is how they
have been used, and here  are the risks.

regards,

Victor K


regards,

Victor K



> _______________________________________________
> v6ops mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/v6ops
>
>