Dynamic DNS (address takeover-type-things)
Tue, 20 Oct 1998 10:56:59 -0400 (EDT)
On Mon, 19 Oct 1998 email@example.com wrote:
> > The first is that DNS is highly cached. That is to say, other DNS servers
> > are caching your answer (albeit for as long as you tell them to) so that you
> > cannot effect an immediate change. Maybe something more like 15 minutes
> > would be effective.
> The difference is that dynamic DNS has protocol extensions to handle this.
That only controls _your_ DNS server. The zillions of others out there are
still going to cache your data for as long as your TTL specifies. And BIND,
at least, enforces a minimum TTL on the data it caches. This is the
authors' attempt at avoiding DNS thrashing I guess.
> > The second is, in your given example (web service) you have a problem in
> > that browsers cache IP addresses practically indefinitely. They are not
> > aware of (or choose to ignore) the TTL specified in the DNS data and cache
> > the IP address as long as they feel like (which could be days in my
> > experience).
> I see my browser constantly doing DNS lookups. It tends to cache pages, but not IP
> addrs (at least on my PC)
Not if you stay on one site. Picture an intranet (or a surfer with very
limited interestes) where the client uses one main server 99% of the time...
> > I guess my point is that an IP fail-over is still a critical part of any
> > complete solution...
> I wasn't trying to replace it, just open up a discussion about what role (if any) this
> technique should have.
I know... I just wanted to make sure we were all on the same page.