[Infrastructures] using IA methodologies to build network element configuration

[Kysh]

On Tue, 29 Mar 2005, Will Lowe wrote:

> > I simply don't know if I agree. I don't want an operating system so
> > much as I want basic control over hardware and firmware, and the
> > logic to handle what I need the box to do.
> 
> Network vendors seem to have enough trouble debugging their code as it
> is; I'd hate to see what happened if the BFR vendors (Foundry, Cisco,
> Juniper, etc) suddenly had to regress not only their own features, but
> bash, vsftpd, apache, and other fairly large, complex open source
> software.

Right, exactly. It should be something that exhibits EXTREMELY 'known'
behaviour; Even if that behaviour isn't always 100% desirable, I rank it
better than something that 'usually' seems to do the right thing. 

Maybe I'm just a minimalist, but the bigger software gets, the less I
want it on something I rely on-- Especially when you can expose a
hundred bugs in just about any normal-looking 15-line chunk of code.

It's the same reason my blood curdles whenever I see Windows on
avionics, whether panel mount or handheld, especially when people rely
on them as their primary source of navigation, or as a 'periphreal
source of navigation' during IFR flight.  Really, is a BFR so different? 

If you can fault my logic, or even if you just disagree with my
approach, I welcome your input. I, for one, am shrinking away from all
the new 'feature rich' stuff that vendors want me to put in the middle
of my network. 

-Kysh
-- 
If this signature were any smaller, it would be smaller.