Rising use of SSL puts pressure on troubled Certificate Authority industry

November 28th, 2012

Secure Sockets Layer (SSL) and its successor, Transport Layer Security (TLS), are the underpinnings of secure online transactions. Using public key encryption (PKI) to encrypt the data sent between a user’s browser and a website, SSL prevents anyone from tampering with the data. Meanwhile the Web server’s digital certificate allows any user to verify the authenticity of the website conducting the transaction. The visual confirmation is the familiar secure lock symbol.

 However, a spate of headline grabbing hacks of certificate authorities exposes security weaknesses in this widely used arrangement. In this LastWatchdog guest post, David Holmes, Technical Marketing Manager at F5 Networks, supplies some context, and discusses the go-forward implications of this major development.

 

Holmes

By David Holmes

The once-cozy, high-margin Certificate Authority (CA) Industry is under attack from three sides. First, the glaring breaches of high-profile CAs, the hacking of Comodo and the complete implosion of DigiNotar in 2011 have brought into question the technical competence of the entire industry. Second, a race-to-the-bottom of quality among new players has resulted in questionable business practices from some of the industry’s established players. Lastly, the external agents are affecting technological changes that the CA industry should have been leading. With so much turmoil swirling about, is there a way forward for the CA industry?

An astute observer may remark that many of these problems have been slowly brewing since 1995 and then ask the question why is this important now? Because the world is converging on SSL, whether it realizes it or not thanks to the ubiquity of malicious actors and the lack of effective international cybercrime law-enforcement. According to a report by Palo Alto, by 2011 23% of all Internet traffic was already using SSL. In the future, a majority of the world’s traffic may be SSL, making reliance on certificates and certificate authorities a fact of life for the foreseeable future

As the world becomes more and more reliant on PKI, non-industry actors such as the Electronic Frontier Foundation, Google (via the Chrome browser), and Moxie Marlinspike have all been pushing their own non-standard technological solutions to problems that should have been solved by leadership within the CA industry. EFF, Firefox and Chrome are “pinning” certificate authorities to domains and Mr. Marlinspike is proposing systems that attempt to undermine Certificate Authorities specifically as brokers of public trust.

Is there a way for the CA industry to regain the high ground and stay relevant? There might be, in the form of participation in the CA/Browser forum. This discussion group was showing signs of life and activity and seemed to offer solutions to many of the industry’s woes, but recently two major CA vendors have pulled out. Is enough being done to save the Certificate Authority industry? Can they be saved from themselves?

The browsers have their own dirty laundry – instead of “failing closed” and denying access to SSL sites when the issuing CA indicates there may a problem with the site, they may connect anyway. Some browsers fail to check the status of the entire certificate chain and support for significant new technologies such as SNI has been hit or miss within the browser community. Neither the CA industry nor the browser community is free of sin.

The Certificate Authority/Browser Forum (henceforth CA/B Forum) is the one place where two sides can come together and to talk. The group communicates via mailing list, regular phone calls, and also meets at physical locations quarterly. Participation is voluntary of course and includes most of the major players that you’d expect. Discussions around unilateral problems such as server latency proceed quickly. Behavioral or policy changes take more time. The output of this group is policy, not standards (that is the IETF’s job).

There is one major technological solution that almost all parties agree would vastly improve the situation, and that is something called “OCSP stapling” which would provide all proper security benefits that a PKI system should confer without the performance penalties that the existing system experiences. Agreement and discussion about OCSP stapling has been encouraging, giving an outsider the impression that the right things are being done, and that the CAs and the browsers may be able to work together after all.

So if there’s a forum where everyone can do their group therapy and sing their campfire songs, is everything good? Well, no, there have been some high-profile defections from the forum recently, including Verizon and Entrust. Entrust had been an active participant (they were the chair) in the discussions but pulled out in advance of intellectual property concerns, which of course, can always be an issue in groups like this. Verizon had been lurking in the forum but hadn’t been active, so the motives behind their retreat aren’t as clear. Some might interpret these defections as signaling trouble within the CA industry itself. A more optimistic interpretation of the situation might suggest that not everyone can be pleased all the time, and that dissension is a natural result of progress. The recent “wins” in the CA/B forum, such as the adoption of baseline requirements, the push for OCSP stapling and the removal of other technological and policy barriers all suggest that the industry is moving in the right direction.

Will it be enough or is it too little too late? Certainly the next two years will be very telling for what the next two decades hold.

 About the essayist. David Holmes  is a spokesperson and evangelist for F5′s security solutions. Since joining F5 in 2001 as a Principal Software Engineer, Holmes has helped design system and core security features of F5′s TMOS operating system. Holmes is a frequent speaker at security conferences such as InfoSec Europe. Prior to joining F5, Holmes served as Vice President of Engineering at Dvorak Development and was a Senior Software Engineer at CyberSafe, Inc.