Re: [apps-discuss] Fwd: Last Call: draft-hammer-hostmeta (Web Host

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Aaron Stone wrote:
> 
> Additionally, the requirements to first check via HTTPS, then via
> HTTP, and the requirements for identical contents, are not
> requirements imposed by RFC 5785 -- though 5785 allows that "a
> registration ... MAY also contain additional information, ... or
> protocol-specific details". A reference to that text might be useful
> to remind an implementer that other well-known URIs may have different
> protocol-specific requirements.

You've found a very serious problem with this document.

The assumption that when a Web-Server is accessible by HTTP on port 80
on some host, then a HTTPS-Server on port 443 will provide access
to the same Web-Server by HTTPS is seriously flawed. 

In the survey here
http://www.esecurityplanet.com/features/article.php/3890171/SSL-Certificates-In-Use-Today-Arent-All-Valid.htm

a simple DNS scan for webservers found 92 million domain names (out of 119)
to host a Web-Server on port 80.  34 (of the 92) millions had an
HTTPS-Server running on port 443 as well.

When performing an SSL-Handshakes on port 443 of these 34 millions
(TLS client without server name indication (TLS Extension SNI)),
only 3.17 percent of these Servers presented a server certificate
matching the hostname used by the client to open the network connection.


In essence this means the recommendation to first try HTTPS, then
HTTP is going to result in ~99% failures to successfully
access the correct Web-Server, and is therefore a _very_ impractical
guidance for an RFC for the real world internet.


Regards,
-Martin
_______________________________________________
Ietf mailing list
Ietf@xxxxxxxx
https://www.ietf.org/mailman/listinfo/ietf


[Index of Archives]     [IETF Annoucements]     [IETF]     [IP Storage]     [Yosemite News]     [Linux SCTP]     [Linux Newbies]     [Fedora Users]