ietf
[Top] [All Lists]

Re: [IETF] Re: We can change the world in a 1000 ways (IPv4 over IPv6)

2013-11-12 13:53:45

On Nov 12, 2013, at 1:12 PM, Michael Richardson 
<mcr+ietf(_at_)sandelman(_dot_)ca> wrote:


Ted Lemon <Ted(_dot_)Lemon(_at_)nominum(_dot_)com> wrote:
It would be nice to convene a summit of operators (at RIPE or NANOG)
and describe the various mechanisms and rather than ask them which one
they like, 
ask them which one they would *NEVER* consider.  That might reduce the
field by half...

I don't think that's practical—they might all vote for a protocol that
they wind up not wanting to deploy.   The models that are under

That's why I wouldn't ask them to pick a winner, but rather to pick the loser.

consideration actually have running code and some operational
experience behind them.   So asking operators to decide based on a
feature list or something of that sort is not a good idea.   What we
really want is for operators who have realistic intentions of deploying
this stuff to weigh in.   And they are doing so, in the working group,
so we don't really need to go to RIPE or NANOG to get this feedback. 

a) I'm not convinced the operators we have are very representative of the
  whole.

It is not, but a number of the operators who do participate *try* and represent 
the industry, and not just their employer.

A well known issue is the lack of operator involvement in the IETF. 
In general operators are focused on solving issues *today* - much of the work 
in the IETF is viewed as much further out  and / or detached from the 
operational real world.
A number of things have reinforced this view, especially some of the v6 
zealotry and unwillingness to listen to what the operators actually want (see 
the DHCP discussions and extension / fragment discussions). Just because it 
works fine in IPv4 is *not* actually a good reason to change it in IPv6...


Very rough stats:
NANOG has ~690 attendees (Phoenix), RIPE has ~570 (Dublin), APRICOT ~725 
(Singapore), lacnic20 and ~291 and MENOG ~95. I was not easily able to find the 
attendee list for other NOGS (like PacNOG, Afrinic, etc.)

There are around 1100 organizations represented (the list formats were not 
identical, some folk spell their organization names differently, do you count 
Globenet and Globenet/Brasil as one organization or two?)


 (But those that show up at a NOG might not be either. sadly)

Yup. But it *is* much more representative...


b) I'm more interested in reasons operators who are not deploying anything,
  have for not wanting to.

I'll bet if we had a single IPv4 over IPv6 solution which had a clear
operating cost savings over Dual-Stack, and also over IPv4-only+CGN, that
we'd be at universal deployment of IPv6 already.

I don't really understand why we have so many mechanisms... Perhaps we could
have an IAB plenary presentation on it... or maybe someone could do an ISOC
video like Kathleen did for MILE.

-- 
Michael Richardson <mcr+IETF(_at_)sandelman(_dot_)ca>, Sandelman Software 
Works 



-- 
Eagles soar but a weasel will never get sucked into a jet engine 


Attachment: signature.asc
Description: Message signed with OpenPGP using GPGMail