SSL/TTS Vulnerability Response (CVE-2009-3555)

Anthony Maughan

There's been much coverage and discussion of recently disclosed SSL vulnerabilities. ANXeBusiness is jumping into the discussion with the help of a special guest blogger, Steve Dispensa. Marsh Ray, who is on Steve's development team, is credited with discovering this vulnerability.

Steve is currently the CTO of PhoneFactor and is one of the founding partners and original developers of our cloud-based remote access product, PositivePRO. The idea behind PositivePRO was that remote access provided by the internal IT staff was expensive and cumbersome to maintain. Steve and his crew were determined to develop a Software-as-a-Service alternative that was simpler and more effective than traditional appliance-based approaches. One of the key objectives was to make it even more secure than traditional standards at the time so that IT Administrators would feel better about using a hosted VPN solution. Steve developed an enhanced encryption method, "WTP" or WebTop Transfer Protocol. Using his own method of encryption in addition to SSL/TTS exchange enhanced security, and because of the strengths and weaknesses of PKI, eliminated the dependencies on the public infrastructure. ANXeBusiness acquired the PositivePro product line last August and has continued to make enhancements to the security and infrastructure originally designed by Steve.

Here's Steve's take on the recent SSL vulnerability disclosures:

"I'd like to thank the ANX team for giving me a chance to tell some of the backstory behind the discovery of this flaw, as well as to discuss what this means for the PositivePRO products.

As we frequently do at PhoneFactor, the team was discussing security architecture in the context of one of our system components last July. I got into a debate with one of our engineers(Marsh Ray) about an aspect of SSL authentication that prompted him to go digging through the source code to our web server. Sometime in August, he emerged having found that all versions of SSL were vulnerable to a trivial man-in-the-middle attack. Not a bad way to spend a couple of weeks!

The flaw has been described in detail elsewhere, including in our original paper, but to summarize, it allows an attacker to get in the middle of any SSL communication and inject whatever data he wants. This is potentially very bad, in that the attacker can cause whatever transaction he wants to be executed.

It's not theoretical any more, either. This flaw was recently leveraged into an attack on Twitter by a Turkish security researcher. The hole has since been patched, but the point has been made: this is a significant vulnerability with hard-to-predict but potentially wide-ranging effects.

So what does all of this mean for PPro? First off, the bulk of connections were never affected by this flaw, since they don't use SSL. For the few areas of the system that did rely on SSL - the web interfaces to Policy Manager and WebTop - patches have been applied to the servers to disable renegotiation, mitigating the attacks.

It's been an interesting experience describing and disclosing this flaw to the industry. Our company has a long history of doing security research and finding significant flaws, but this is by far the largest-scale flaw we've ever reported. We went to an industry consortium of core IT vendors called ICASI, along with OpenSSL, Apache, the NSS team (the SSL library in Firefox), and representatives from the IETF, to describe the flaw during an in-person meeting at the headquarters of Google in Mountain View, CA, on September 29th. At that point, we thought we had a few months to work out a solution and implement it. You can imagine my surprise when I saw an e-mail posted to a public IETF mailing list with the exact problem described in the subject line!

Going forward, we expect most vendors to release patches soon. Work continues in the IETF to get to consensus on a permanent solution to the problem, but deployment of a new standard could take years. Meanwhile, I encourage administrators to be proactive in deploying mitigations, including applying vendor-supplied patches, running IDS's and firewalls, and of course, making sure all remote access is secured with a VPN like PositivePRO and a two-factor authentication system like PhoneFactor!"