Re: [TLS] Eleven out of every ten SSL certs aren't valid

[Ivan Ristic <[e-mail]>]

The numbers in the article come from the preview of my research, which
I presented in a webcast last Thursday. Here's the summary of the
presentation:

1. I started with 119M domain names (out of 193M registered
worldwide). The 119M include all .com, .org, .net, .info, .biz, and
.us domain names.
2. 92M domains are active on port 80 or port 443
3. 33M domains have port 443 open
4. 22.65M domain names run SSL on port 443
5. On 0.72M domain names certificates match the domain name.

I am now focusing on the 720K certificates that are potentially valid.

Once I find out exactly how many of those certificates are valid, I
will make two claims:

- X% of web servers properly run SSL (where X will be < 3%)
- Y% of certificates are valid (where Y will be significantly higher than > 3%)

I should be able to post my presentation online tomorrow, after which
I will follow-up here.

BTW, my assessment methodology is available on
https://www.ssllabs.com, along with an online assessment tool (that
works with a single hostname). There's additional material on the SSL
Labs mailing list. I will publish the complete report in about a
month.

I welcome all feedback, as well as deeper involvement if you're
interested in the topic.

Cheers,
Ivan

On Tue, Jun 29, 2010 at 8:50 AM, Peter Gutmann
<[e-mail]> wrote:
> In case someone here still hasn't seen this, the subject is a reference to:
>
>  SSL Certificates In Use Today Aren't All Valid
>  http://www.esecurityplanet.com/features/article.php/3890171/SSL-Certificates-In-Use-Today-Arent-All-Valid.htm
>
> which posits that only 3% of SSL certs in use today are valid.  The figures
> seem a bit suspicious though, for example they claim 23 million SSL sites
> while the same article quotes Netcraft as claiming there are 1.5 million SSL
> certs in use (the Netcraft figures may be for CA-issued certs only, since they
> quote Verisign as a percentage of that total).  Still, 3% seems pretty low,
> could this be due to something like virtual hosting and the client not sending
> the hostname, thereby getting the wrong cert?  Even with that though, I
> wouldn't have expected a 97% invalidity rate.
>
> Peter.
> _______________________________________________
> TLS mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/tls
>

-- 
Ivan Ristic
ModSecurity Handbook [http://www.modsecurityhandbook.com]
SSL Labs [https://www.ssllabs.com/ssldb/]