There’s a lot of guidance needed for operators and tools developers (who set 
defaults) with regards to DNSSEC.  (I.e., Define and quantify “useless.”)

Three old presentations on studies (involving operations TLDs only) I have done 
on this topic in 2012:

http://iepg.org/2012-03-ietf83/IETF83IEPGLewis.pdf
(how the protocol engineer’s expectations differed from what operators were 
doing)

https://ripe64.ripe.net/presentations/46-RIPE64PlenaryTLDDNSSEC.pdf
(began to analyze the difference between protocol engineering and operations)

https://www.nanog.org/meetings/nanog56/presentations/Tuesday/tues.general.lewis.14.pdf
(same, based on more data)

If you are going to produce such a document, cover a range of choices being 
made, including the salt change frequency.  There’s been a strong need for a 
DNSSEC implementers guide covering cryptographic parameter selection ever since 
well, forever.  Yes we have NIST 800-81 
(http://csrc.nist.gov/publications/nistpubs/800-81r1/sp-800-81r1.pdf).

One issue with the salt (in particular) is that the RFC recommendation is (RFC 
5155  Appendix C.1) says this:

"The salt SHOULD be changed periodically to prevent pre-computation using a 
single salt.  It is RECOMMENDED that the salt be changed for every re-signing."

This was written without dynamically updated zones in mind (or so I’m guessing 
- 5155 is “not my fault” ;) ).

On Jul 24, 2014, at 18:02, Mark Andrews <[email protected]> wrote:

> 
> I just sent the following to bind-users.  We need to kill the myth
> that changing NSEC3 salt provides any real benefit.
> 
> "Actually it is useless to change the salt regularly.  Changing the
> salt provides no real benefit against discovering the names in a
> zone which is the reason people were saying to change the salt.
> 
> The attacker uses cached NSEC3 records.  When it gets a cache miss
> it asks the servers for the zone, puts the answer in the cache and
> continues.  When the salt changes it just maintains multiple nsec3
> chains eventually discarding the old nsec3 chain eventually.  I
> would wait until the new NSEC3 chain has as many cached records as
> the old NSEC3 chain.  Changing the salt slows things up miniminally
> for a very short period of time after the change.  Additionally
> once you have some names you ask for those names for a non-exisisting
> type to quickly pull in part of the new NSEC3 chain you know exists.
> 
> The only reason to change the salt is if you have a collision of
> the hashed names.  This will be a very very very rare event."
> 
> Mark
> -- 
> Mark Andrews, ISC
> 1 Seymour St., Dundas Valley, NSW 2117, Australia
> PHONE:        +61 2 9871 4742                  INTERNET: [email protected]
> 
> _______________________________________________
> DNSOP mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/dnsop

_______________________________________________
DNSOP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dnsop

Reply via email to