>From: sidr [mailto:[email protected]] On Behalf Of Demian Rosenkranz
>
>Correct, Im talking about really short lifetimes for ROAs (EE certificates). 
>The
>RP software would be forced to cryptographically checks the objects again
>and again over short intervals.
>But long lifetimes for ROAs (EE certificates) mean at least bigger CRLs.
>This would be one benefit of short lifetimes.
>

Even if about a few hundred origination-change events occur in a year that 
require 
ROA-EE-certificate rollover, you are dealing with an increase of merely just 
that many 
additional entries in the CRL (with the long-lifetime ROAs and revocation 
approach).
If instead short lifetimes are used, then 500,000 certs and ROAs would be 
propagated
in the RPKI system periodically in each of those short intervals. 
The latter seems to be a much bigger price to pay.
But if you can provide further analysis and insight in your thesis, 
it would be very welcome.

We discussed and quantified these types choices and trade-offs earlier 
not in the context of ROA (EE cert) lifetimes, but in the context 
of AS or router key rollover mechanisms to mitigate BGPSEC update replay 
attacks.
Please see:
http://tools.ietf.org/html/draft-sriram-replay-protection-design-discussion-02 
http://www.ietf.org/proceedings/85/slides/slides-85-sidr-4.pdf
http://tools.ietf.org/html/draft-ietf-sidr-bgpsec-rollover-02 

Sriram

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

Reply via email to