Re: how to protect name servers against cache corruption

[Ben Black <black () zen cypher net>]

i say again that although it cannot be made completely secure in the 
DNSSEC sense, it can absolutely be made far more resistant to some 
*known* attacks without significant code changes.


ben



On Tue, 29 Jul 1997, Paul A Vixie wrote:

> Let me put this another more interesting and more direct way.
>
> Postulate a name server with the following properties:
>
>       1. Actually works on and is connected to the live Internet.
>       2. RFC compliant except as nec'y to comply with #1 above.
>       3. No DNSSEC, no TSIG, no SECUPD.
>       4. Completely bug free.
>
> You go right ahead and build that name server, and I will drive a truck,
> no, better still a bus or even a backhoe, right through its front window.
>
> DNS is not secure and cannot be made so.  BIND-8.1.1 is the best there is,
> and it's what you should run, but as long as you run DNS without DNSSEC,
> your confidence level should be set accordingly.
>
> PS:
>
> BIND is definitely #1, is almost #2, is definitely #3, and trying to be #4.