Re: [Dance] CRLs/OCSP and DANE at RIPE84

As I said in the presentation, it still seems remarkably wrong that the 
X.509 certificate system handles revocation so badly. Now this would not be
a problem if Evil Acts took more than the average X.509 certificate lifetime 
to set up and execute, but given that it takes only a small number of hours
for a key compromise to be exploited, then the conclusion is that this 
X.509 framework is simply not fit for purpose.

I appreciate the depressing reality that saying that X.509 is the massive incumbent
and we’ve grown used to its foibles and weaknesses, but its about as useful for
victims of certificate malfeasance as a full armour plating made of wet lettuce
leaves! Certificate lifetimes are working on scales of months and years, revocation
is broken and can't scale, and bad practices and evil deeds are executed
in hours or even minutes. The mismatch of attack and defence capabilities
plays in favour of the attacker.

I know we’ve tilted at this X.509 windmill for many years, but the observation
I would make is that in no case as the X.509 environment changed in response.
The exposed problems appear to be a constant. On the other hand attackers
do learn and adapt and the exploited vulnerabilities in the very design
of the X.509 system become factors every just has to accommodate. 

I am still optimistic about DANE and DANCE. To my mind there is little else
around that can make this sad state of affairs any better.

Sorry for taking up the WG’s time here. I know this is straying off topic!

Geoff

> On 24 May 2022, at 12:33 am, Shumon Huque <shuque@gmail.com> wrote:
> 
> On Sat, May 21, 2022 at 7:18 AM Michael Richardson <mcr+ietf@sandelman.ca> wrote:
> 
> https://ripe84.ripe.net/archives/video/864/
> Geoff Houston looks at Revocation, and who it is just not working, and suggests DNSSEC+DANE.
> Very much Worth watching.
> 
> I'm kind of sympathetic to Geoff's views.
> 
> But the prospects of DANE generally replacing (or constraining) PKIX and delivering a DNS
> based revocation capability seem pretty slim to me, especially in the web arena, which seemed
> to be the focus of Geoff's presentation. Note the failed attempt to standardize the TLS DNSSEC
> chain extension in the TLS working group (now published as an experimental RFC via the ISE).
> 
> In other application areas, like DANCE, there will hopefully be better prospects.
> 
> Shumon.
> 
> -- 
> Dance mailing list
> Dance@ietf.org
> https://www.ietf.org/mailman/listinfo/dance