Re: focussing energies

[Pekka Savola <pekkas@netcore.fi>]

This is a very important topic, but as it mostly goes to issues beyond 
our control, let's hope the thread does not get out of hand...

On Mon, 15 Mar 2004, Alain Durand wrote:
[...]
> b) assisted tunneling works best in the scenario of an ISP willing
> to offer IPv6 service but not ready yet to pay to deploy native
> service. Those mechanisms, like tunnel broker, not only provide IPv6
> connectivity to the user, but also to help the ISP to jump start
> IPv6 service to its customers at low cost. This is an area where
> standardization from this wg could help a lot.
> 
> c) when ISPs are not cooperating, there is the choice between two evils:
> - fully automatic solutions, with their share of complexity, security 
> issues,...
> - tunnel brokers operated by third parties, with possible sub-optimal 
> paths and complex set-up.

[...]

> The point I'm trying to make is that, IMHO, this wg should not spend to 
> much effort on c)
> [i.e. there are implemented solution, let's document them and move on]
> because:
> 	1) either IPv6 will take off and ISPs will start to cooperate, thus 
> we're back to case b)
> 	2) ISP still don't cooperate in the near future, meaning that they see 
> IPv6 as going nowhere,
> 	so why should this wg and software vendors invest in complex
> transition mechanisms? 

There is a middle ground here: ISPs not knowing when is the right take
to start taking IPv6 seriously, and actually deploying it in a
fast-track fashion.  Most, especially those who look the red/black
bottom line, have been just biding their time.  They wait for the user
demand or some push from any direction.  This push could be achieved
either through a) [which is outside of our scope] or through some v6
application getting popular.  Barring outside agencies, we seem to
need to work on enabling that "application landslide".  More of that
below.

> and focus its energies on standardizing a
> solution for b)

My take on this is that being able to crack the chicken-and-egg
problem is likely to take a while -- measured in years.  We have to
provide means for app developers to deploy v6 applications which might
make IPv6 fly.  In the general case, deploying v6 applications is only
possible if there is sufficient IPv6 penetration, e.g., through the
use of automatic transition mechanisms.  (Otherwise the app developers
just keep using IPv4 band-aids, and we never get to IPv6 in the first
place -- and similarly, if an app is v4-capable, the users/vendors
don't see the push to move to v6!)

So, my personal gut feeling is that we need to tackle the first
subcase of c), while the second is not useful for the mass deployment.
If all goes well, this stage would be only short-term.

b) is equally interesting; if coupled with a) in particular, the local
ISPs might be keen to offer some IPv6 service.  This is probably
something that could be applicable for a slighly longer term than c)  
subcase 1).

-- 
Pekka Savola                 "You each name yourselves king, yet the
Netcore Oy                    kingdom bleeds."
Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings