Latency

[Erik Guttman]

On Mon, 3 Feb 2003, Spencer Dawkins wrote:
 
> I agree with most of what Henning is saying, but am disturbed by one
> point -
>
> > [...] Almost any working group 
> > has to now 
> > go through a much longer chartering process, fairly extensive 
> > 'requirements' and 'framework' documents, before the working group is 
> > even allowed to talk about protocol development. This is a process 
> > change which was (I believe) never really discussed and isn't really 
> > documented.
> 
> How widespread is this perception? I have no idea WHY individuals in
> the IETF would think they were "not even allowed to talk about
> protocol development" before they go "extensive 'requirements' and
> 'framework' documents". 

Many WGs are required to put together requirement documents before they
are chartered to begin work on protocols.  AAA was an extreme example of
this.  While it worked for AAA (to decide which technical direction to
embark on among competitors) it incurred all the costs Henning listed.

An example where an architecture and protocols were developed without a
definitive requirements spec is instructive:  IPsec.  Contention was not
dealt with up front, which led to a longer process of protocol selection
and difficulty determining 'completion criteria.' This effort did not
follow a waterfall model so much as a waterspout, with powerful forces
pulling in many unchartable directions. 

Moral:  Complex, contentious protocols with nuanced security aspects
take a lot of time to produce.  

I think this argues for humility and projects with bounded scope.  These 
can be built upon by future work if designed with that as a central
objective.

Erik