Mysterious missing DHCPv6 feature, was Re: How does one obtain an IPv6 DNS server when VPNing to an ASA?

[David Barak]

--- On Tue, 5/18/10, Benedikt Stockebrand <me at benedikt-stockebrand.de> wrote:

> Any change to the existing Internet, especially if its as
> "radical" as
> IPv6, is a disadvantage to the organizations "which have
> made
> significant infrastructure investments in this".
> 
> This is especially so because the Internet *is* changing,
> and if those
> organizations believe they can stop these changes, they'll
> be overrun
> by a new generation of better-value-for-less-money
> providers.

To paraphrase Randy Bush, I encourage my competition to take this approach to service deployment.  

> 
> Tough, but that's the way how business in a free market
> works.
> 
> > Are you honestly suggesting that this should be
> swallowed without
> > comment?
> 
> "Comment" is ok...
> 
> > What if the design decisions are in fact bad? 
> 
> ... and so is asking these questions---in the right place
> ...
> 
> > Why not try to fix the brokenness?
> 
> ... but showing up after ten or more years complaining that
> one's
> existing business model isn't protected is not.
> IPv6 deployment is long overdue.  Coming up with
> reasonings "but we
> can do this slightly better if we do it another way"
> however is pretty
> embarrassing at best.

Some of us have been complaining about the brokenness of the IPv6 autoconf approach for a long time.  Check your archives, and you'll see that this is true.  For years, the response was "RA is sufficient, and will make DHCP obsolete."  Now that DHCPv6 is becoming a requirement, it is only a matter of time before every feature present in DHCPv4 is also present in DHCPv6.  This isn't about "protecting a business model" - it's about making very large scale deployments work *at scale*.  There is a *currently working* approach.  The burden of proof is on those who want things to be different, not on those who want it to be the same.

If you don't understand the difference between the complexity of a single centralized management server and that of a centralized management server plus N midpoints (where N ~= number of customers), then nothing I say is going to help: I encourage rereading RFC 1925.

But hey, if you think that's embarrassing, that's fine too.

David Barak
Need Geek Rock?  Try The Franchise: 
http://www.listentothefranchise.com