IPv6 dynamic DNS services?

Benedikt Stockebrand me at benedikt-stockebrand.de
Thu Mar 17 11:33:12 CET 2011


Hi Daniel and list,

Daniel Roesen <dr at cluenet.de> writes:

> Also keep in mind that this usually operates in a single DNS domain
> so you need per-RR authorization controls (BIND e.g. doesn't have
> that as far as I'm aware - I might be wrong).

BIND actually has, but last time I messed around with it I wound up
writing into a file included by named.conf (I'm not sure if that could
have been avoided, though).

In any case it is possible to allow TSIGs to update only a given name,
or a given RR type for a given name.  For a lab grade/proof of concept
implementation, check out

    http://www.benedikt-stockebrand.de/maketsigkey-1.0.2.tar.gz
    http://www.benedikt-stockebrand.de/nsautoupdate-1.0.1.tar.gz

> Implementing a DNS UPDATE + TSIG client is a bigger effort than
> just the simple HTTP interfaces currently being used which just needs
> a slight modification.

Depends.

If you use a Un*x system, the nsautoupdate script mentioned above uses
the nsupdate binary included with BIND, or rather the BIND client side
binaries; so the implementation effort that way is negligible.

If you implement it in embedded systems, SSL should be *significantly*
more complex/costly than TSIG at least unless the embedded system
already includes SSL support for a web interface or such.  Even
without strong crypto, HTTP/TCP is more complex than DNS/UDP.  (And
comparing DNS+TSIG/UDP with non-SSL HTTP/TCP isn't exactly fair.)


Cheers,

    Benedikt

-- 
			 Business Grade IPv6
		    Consulting, Training, Projects

Benedikt Stockebrand, Dipl.-Inform.   http://www.benedikt-stockebrand.de/



More information about the ipv6-ops mailing list