[nycbug-talk] IPv6 Followup

Adam Rothschild asr+nycbug at latency.net
Sun Nov 11 20:35:44 EST 2007


On 2007-11-11-19:25:33, Isaac Levy <ike at lesmuug.org> wrote:
> And I'll add to that, look past the glossy IT-industry mags, and
> look over at the the IIJ...
                       ^^^

For a peak at operational reality as it relates to v6 adaptation, one
would be well-served to review these slides from the recent NANOG
conference:

  http://www.nanog.org/mtg-0710/presentations/Bush-v6-op-reality.pdf

For those unaware, Randy is a top scientist at IIJ, reporting straight
to the CEO, and has been around the block long enough to know a thing
or two about v6.

On 2007-11-11-19:14:13, Isaac Levy <ike at lesmuug.org> wrote:
> Look Alex- could you please stop derailing every thread that comes
> up on IPv6? 
[...]
> In no uncertan terms- cut it out.

Unfortunately, things aren't as black-and-white as you make them out
to be.  Things are not rosy, delivering v6 services to customers is
not as simple as it may seem from an end user's prospective, and at
the end of the day, demand from revenue-generating customers (versus
hobbyists) is simply not there.  I, for one, welcome Alex's fresh
breath of reality in these discussions, and would be disappointed
should you choose to censor his dissenting opinions in the interest of
maintaining the peace.

> - If you had come to one of the last 2 NYC*BUG meetings on the  
> subject, you'd have had ample opportunity to see people show up who  
> care about IPv6.

Nobody's doubting that the interest is there.

Unfortunately, little to no interest is coming from large enterprises,
content providers, nor access providers -- typically the largest
sources of traffic and revenue for tier-1-ish ISPs. 

> Sixxs and HE are the only mass-consumption options available, and
> until people in your position (and larger) start selling usable IPv6
> pipes, there will be even more tunnel use...

It's easy to beat up Alex (or anybody in his shoes really) for not
implementing v6.  Perhaps a more productive course of action would be
to ask "why not".

Lack of demand (see above) plays a key factor here.

Even if the demand were there, purchasing v6 transit on a large scale
isn't exactly simple.  With the exception of NTT/Verio (AS 2914), I'm
unaware of any tier 1 providers running dual-stack.  Tunneling is
ghetto, doesn't scale from a traffic prospective, and often times
provides the user with a worse-than-v4 browsing experience given
divergent physical and logical topologies.  Even the folk lobbying in
favor of v6 at every single conference (*cough* 701) are ill-prepared
to actually offer it to (paying) transit customers.

> - You are not the arbiter of meaningful use of the Internet.

Nor are, unfortunately, you.  While Alex might not deliver good news,
I would think it unwise to shoot the messenger...

-a



More information about the talk mailing list