[CII] terms and conditions

[bmanning at vacation.karoshi.com]

On Fri, Nov 27, 2009 at 10:36:55PM -0700, John Osmon wrote:
> On Fri, Nov 27, 2009 at 10:24:46PM +0000, bmanning at vacation.karoshi.com wrote:
> > 
> > well, well...
> > 
> > 
> > 	Taking each word in turn:
> 
> [...definitions elided...]
>  
> > Perhaps all are true to a degree, but I think it would help if we were
> > to settle on one or at least be clear when we are talking,
> > just which things we talk about.
> > 
> > What do you all think?
> 
> Wit all respect for Justice Stewart:
>    I don't know what the Internet is, but I know when I see something
>    crtical.  :-)
> 
> So far, people have been happy to point out that they aren't bots, 
> and they they would like to use this list to learn what others think
> is critical. 
> 
> me?  I want to see IP networks beome so ubiquitious and so nimble
> that they can be used for any critical necessity.  This means that
> any given network will need to have flexible policies  -- at different
> times, different traffic will become critical.
> 
> I don't think the typical view of "critical" is able to deal with 
> what I think networks should be able to handle.  We still live in
> a world where too many people tie the end application to the network
> link.  Technology moves forward faster than regulation can keep up...
> 
> "My network, my rules."  But I'll always prempt traffic on my net
> for health/safety.  Kinda seems like my duty to the society that I live
> within...
> 
> We can argure tighter semantics once other people start talking
> about what they consider "critical."


	one of the often misunderstood attributes of an IP network
	is the basic attribute of the End2End principle.  e.g.
	the network can and perhaps should be decomposed on a periodic
	basis...  there is nothing sacrosanct about any given IP network
	or set of interconnects between any set of IP networks.  

	this is not much more than normal network continuity/disaster testing -
	I mean we periodically test out backup power, hotsite testing, etc...
	why don't we test out backup peering paths?  BGP is pairwise, so it
	works well.

	as to tying applications to the network... we are kind of stuck
	with the DNS and the one unique root...  Now this could be fixed - 
	without much change in the namespace or even the authoritative servers.
	(send me a note if you want to know more ...)

--bill