[arin-ppml] IPv6 Non-connected networks

[William Herrin]

On Fri, Feb 5, 2010 at 1:25 AM, George Bonser <gbonser at seven.com> wrote:
>> My point was that I'd really like to see us stop suffering failures of
>> imagination in the name of careful reason. ARIN has to be the
>> moderating force on the routing table only because we've failed to
>> imagine a individually driven process that renders the role
>> unnecessary.
>
> And my point was that the size of the routing table is only an issue
> because of hardware vendors lack of resources in their gear.  Having
> more resources available eliminates the entire issue.

Well sure. But are you familiar with how the hardware works in the
high-end routers? The stuff is seriously expensive to build,
especially in the relatively low quantities that populate the core of
the DFZ.


> A disruptive
> player in the hardware market that introduces a router that can handle
> two or four or eight times the number of concurrent flows and routes as
> current gear would moot the point.

The current kit deals with packets rather than flows. Flows was
investigated thoroughly and looks very much like a blind alley. It has
some useful niches but the failure modes are catastrophic.

Granted someone could stumble on a brilliant insight tomorrow,
especially if under pressure to innovate,. But understand that you're
talking about a scientific advance, not a technological advance.
Technology can be predictably built from the existing science at
demand. Scientific advances are unpredictable; they happen almost at
random. If you're a betting man, don't bet on the timely discovery of
science that doesn't yet exist and right now the science of routing is
Tries or TCAMs.

Regards,
Bill Herrin


-- 
William D. Herrin ................ herrin at dirtside.com  bill at herrin.us
3005 Crane Dr. ...................... Web: <http://bill.herrin.us/>
Falls Church, VA 22042-3004