Re: [idn] Requirements I-D

["Brian W. Spolarich" <briansp@acm.org>]

On Fri, 19 May 2000, Harald Tveit Alvestrand wrote:

| you make a convincing argument that we need more than one "DNS"; the
| DNS as presently constituted can't do all that and do its other jobs
| at the same time, I think.

  I'm not sure if my vague note constitutes a convincing argument, but I
have been wondering about this a bit as I've been lurking.

  The requirements as outlined in the I-D are pretty daunting, especially
from the compatibility standpoint.  One might be tempted to look at the
problem and say "This compatibility stuff is going to make the solution
look really ugly.  Lets just start over."  This would imply an IDNS
protocol development exercise, with an implementation that would run on a
new well-known port and would require implementation on every platform
that currently runs a resolver.  I think that the widespread adoption of
the incremental approach that's currently being considered is going to
take quite some time to become widely-deployed once an implementation is
available, which will be quite some time from now, given where we are in
this process.  A 'greenfields' approach would take forever to build enough
consensus around, and would likely be deployed sometime after IPv6 is
finally in place. :-)  I hope to be retired by then.

  I think that one of the great strengths of the DNS as is currently
implemented is its relative simplicity.  The number of protocol operations
is relatively small, the concepts are relatively easy to understand, and
the interface is simple for novice programmers to use properly.  For
administrators, BIND is easy to setup and administer, although it is easy
to make mistakes that cause unexpected results.  This in the context of
what is arguably the world's largest distributed database.

  In defining a picture of the IDNS, I think its important that this
approach of practicality and simplicity be kept in mind.  Perhaps as a
followup to my previous comments about 'What Is the DNS' it would be
helpful to frame some context in the I-D about how the DNS is used today?

  -bws

-- 
 Brian W. Spolarich - Ann Arbor, MI - briansp@acm.org - www.acm.org/~briansp/
        "Wickedness is a myth invented by good people to account for
            the curious attractiveness of others." - Oscar Wilde