ietf
[Top] [All Lists]

Re: utility of dynamic DNS

2002-02-28 13:50:04
It's also quite clear to me that stable DNS names are not an
adequate substitute for stable IP addresses , and that the existence
of a service that can be used to update DNS names when IP addresses
change should not be taken as an indication (for example) that it's
okay for providers to change IP addresses at a whim...

I'm not sure I understand this. For most of the application layer
protocols that I deal with everyday (SMTP, POP, IMAP, HTTP) , and a
large class of similar protocols, a stable DNS name is perfectly
reasonable substitute for a stable IP address. Do you think that
there is not a huge class of protocols that can reasonably use
dynamic DNS, or do you simply think that there are a significant
number of problems for which a stable IP address is required?

Several popular protocols that meet certain criteria - the connection 
durations are short, etc - can indeed do okay with relatively unstable 
addresses as long as the name-to-address binding is kept current.
But this doesn't generalize to all or even most protocols, or even to 
all uses of the protocols you've mentioned.  

For instance, if you need to transfer large files over a slow/busy link 
via HTTP (or FTP) such that a file might require several hours to 
transfer, you may find that the connections break too often for dynamic 
DNS to solve the problem.  The same would be true for long email messages 
over SMTP.

Again, it's not that I have an inherent beef with Dynamic DNS. 
I just disagree that DNS names are suitable as a general substitute 
for stable IP addresses.

Keith