Re: [Int-area] Review of draft-narten-ipv6-3177bis-48boundary-05

[Mark Smith <[e-mail]>]

Hi Fred, Tony,

On Fri, 20 Aug 2010 13:16:05 -0700
Fred Baker <[e-mail]> wrote:

> Thanks, Tony.
> 
> Let me comment on one point in your review.
> 
> On Aug 20, 2010, at 11:47 AM, Tony Li wrote:
> 
<snip>
> 
> To be really honest, I have concluded that every time we further idiot-proof the world, the world makes better idiots.
> 

(That made me laugh out loud when I first read it :-) )

That seems to me to be a good argument for less options and more
simplicity. The less knobs there are to tweak, the less chance of them
being tweaked incorrectly.

This topic has made me think a bit more about why I've liked e.g.
Novell's IPX, and why I think IPv6 should be at least as easy to use.

Protocols like IPX and Appletalk were easier to use because they were
designed to be user friendly - with the users of the protocols being
both the end users and the operators of the network (if the network
was large enough for them to exist). User friendliness is quite an
attractive feature of a protocol once you've experienced it.

I think the main reason I've liked the idea of everybody having a /48
as a minimum is that it creates more simplicity and therefore user (and
operator) friendliness. It's one less thing that varies, and therefore
one less thing that the better idiots are likely to get wrong. It also
makes my life easier as an operator because it's one less parameter to
track the value of when dealing with the majority of addressing tasks.

Having a "one size fits all" approach to addressing has worked
successfully with ethernet addressing and other layer 3 protocols, so I
think the principle is quite sound. The cost has been more
than functionally necessary addressing bits, however those bits are
generally pretty cheap and have provided a good simplicity and
convenience return.

So while I'm not strongly against the idea of having a small variety of
prefix lengths (which is a whole lot better than the 20+ in IPv4!), I
think it is worth trying to pursue operational simplicity and
convenience, as a priority, by fixing parameters that don't need to
vary. Hopefully that will continue to raise the threshold of the
quality of idiots, and the efforts they have to go to, to be successful
at breaking things :-) I think most low quality idiots won't bother when
things just work :-) 

I'd really like to see IPv6 being far more user friendly than IPv4 and
at least as user friendly as more modern protocols like IPX and
Appletalk. I think the original designs and parameters pursued that
goal, however as the protocol has been exposed to more and more
IPv4-only people, it seems to me that there is now a push to make it
more like IPv4 because those people are comfortable with IPv4 and
haven't experienced the user friendliness of anything else. I think this
is ultimately going to make IPv6 less user friendly than it was designed
to be and can be. That'd be a shame.

Regards,
Mark.