I would offer that we select the "thing" that looks the most persistent
to be the persistent identity. If the choices are: DNS name vs IP
address, I think most people would recognize that the DNS name is the
persistent identity. And it is probably the one most people would want
to use, especially in light of the problem of writing down IP addresses.
This is just when we are talking about Hosts. What if we were to
include services?
In terms of scalability debates, I have yet to see an analysis to
support claims of more or less scalable wrt DDNS, Mobile IP, etc. We
should probably try to move the debate from "proof by emphatic
assertion" to analysis.
Most of the statements I have seen on the scalability topic remind of
the debates in prior decades used by a couple of telcos to dismiss the
viability of mobile cellular telephony. In the end, the HLR/VLR
paradigm has been one of the most successful deployments in the history
of technology (the wheel has it beat, not clear the Internet has - yet).
(and I can't help mentioning that HLR/VLR is an example of an
architecture analogous to DDNS where a mobile phone number is a name).
Perhaps there are theses already tackling these issues?
At a minimum, DDNS covers today's common cases of managing the DNS.
It's utility is quite good for the common roaming/mobile cases, today.
More importantly, it is used to solve the basic network Name to IP
binding problem for common deployments such as enterprise. Everyday
people move across IP subnets due to office moves and DDNS just works
without network administration intervention.
In reality, the debate should not end up being either/or in nature. It
is likely that a multi-tier set of mechanims will operate. We already
see that with services where we have very long lived names in the DNS,
shorter lived names in the IP space, and short term mapping of those IP
addresses to actual machines being managed by load balancers and their
ilk.
Regards, peter
-----Original Message-----
From: Geoff Huston [mailto:gih(_at_)telstra(_dot_)net]
Sent: Friday, March 01, 2002 2:04 AM
To: Francis Dupont
Cc: Theodore Tso; Keith Moore; Pete Resnick; Randy Bush; Jakob Schlyter;
ietf(_at_)ietf(_dot_)org
Subject: Re: utility of dynamic DNS
The essence of the architecture of mobility is to allow the identity of
the
mobile device to remain constant while allowing the identity of the
location of the device within the network to vary. The dynamic DNS
approach attempts to bind the domain name as the device's persistent
identity and allows the current IP address to equate to the device's
current location.
Obviously, as already pointed out, the restriction here is that the
device
cannot support persistent state across location changes, but worse, as
far
as I can tell, is that it is an approach that has poor scaling
properties.
In order to operate correctly in a timely fashion the relevant parts of
the
DNS now require very short TTLs. At that point many of the assumptions
of
the scaleability of the DNS tend to be called into question. Is the gain
worth the potential scaling pain?
If the issue here is one of circumventing some level of circuitous
traffic
paths, then it seems to me that there is really not much to gain - any
cursory examination of Internet paths in terms of cable route miles
reveals
a healthy level of gratuitous overhead in any case - the so called
direct
paths you get from dynamic DNS updates as a solution to mobility may
well
be no better than the dog-leg route you were attempting to avoid.