ietf
[Top] [All Lists]

Re: draft-ietf-nat-protocol-complications-02.txt

2000-04-26 00:20:02
From: Bill Manning <bmanning(_at_)ISI(_dot_)EDU>
So, of the 7763 visable servers, 45 are improperly configured in the
visable US. tree. Thats 4.53% of those servers being "not well
maintained.

Keith, These two data points seem to bear your assertion out.

It is always possible to do something poorly.  You can take the best
engineered product and misconfigure it, or otherwise not maintain it.
Sorta like not changing the oil.  On the other hand, how many times do you
see a name service failure for someone in the Keynote 40?  How often do you
get a failure for a site that uses a commercial service like Akamai?  If
you still get lots of failures when people are using the mechanism as
recommended, then the mechanism itself is poorly designed.  It also strikes
me that DNS is continuing to mature, and that things are getting better, as
more products come to the market, but this is more gut feel.

There are some open questions out there: does DNS provide sufficient
granularity?  Are its semantics rich enough for a mobile world?  People are
currently playing tricks with DNS that PVM had not envisioned, and Paul
Vixie loves to say something like, "DNS is not the droid you're looking
for" (forgive me, Paul, if I didn't get that quite right).

Just some thoughts...