Dynamic IP & cache DNS

[Kevin Darcy]

Cricket Liu wrote:

> > So TTL values plummet, and the query volume and the loads on nameservers
> > keeps going up and up and up.
>
> In my experience, most folks have enough spare capacity on their name
> servers
> so that several times the query volume wouldn't be a problem.  There are
> some exceptions, including the root and gTLD name servers, but these aren't
> affected by low TTLs on RRs in zones further down the tree.  I suppose you
> could argue that the big, caching-only name servers run by ISPs might have a
> problem if lower TTLs became pervasive, but that's about it.

Well, waste is waste, and all it takes is for the rate of increase in query
volume, exacerbated by falling TTL values, to slightly outpace the rate of
increase of network bandwidth and/or machine capacity, and the problem becomes
inevitable. I'd rather act pre-emptively now than *re*act when the problem is
in full swing.

Besides, isn't it just common sense to put a little more intelligence at both
ends of a network pipe, rather than pollute the pipe with useless redundant
queries? That is, after all, the rationale for making caching integral to
DNS in the first place. The fact that so many folks are feeling it necessary to
defeat traditional caching, or drastically reduce its effect, just to eke out
some dynamism, tells me it might be time for the DNS caching model to evolve to
the next level.


- Kevin