On Fri, 19 Nov 2004, Charles Duffy wrote:

> On Fri, 19 Nov 2004 07:52:33 -0500, Michael Scheidell wrote:
> 
> > An application inspection firewall like a CyberGuard would (I hope?)
> > detect the non HTTPS type traffic on port 443 or none HTTP traffic on
> > port 80 (Anyone with a CyberGuard care to comment?)
> 
> Once an HTTPS connection is created, or a valid HTTP GET/POST request
> sent, one can still put arbitrary data in as... well, data. Unless the
> software detects and stops you from sending or receiving random-looking
> data streams as files being retrieved by HTTP, someone willing to write a
> little code can pretty easily tunnel their VPN through entirely valid
> HTTP(S) traffic.

While OpenVPN can use TCP port 443 or tunnel over a proxy using the HTTP 
CONNECT method, it makes no effort to impersonate the HTTP or HTTPS 
protocols.

So any proxy that sanity-checks the HTTP CONNECT clients to make sure they 
are talking real HTTPS would be able to block OpenVPN.

Now of course, that doesn't mean that someone couldn't develop a stealth 
patch to talk true HTTP or HTTPS and transmit the tunnel payload using 
GET/POST.

James
_______________________________________________
Nessus mailing list
[EMAIL PROTECTED]
http://mail.nessus.org/mailman/listinfo/nessus

Reply via email to