Revenge of the PKI Nerds

I admit I have a certain fondness for Public Key Infrastructure, or PKI as it is commonly knownat least that is the three-letter version. PKI is commonly described using choice four-letter words as well. That’s because it came into favorand just as ingloriously fell out of itwith the boom of the ’90s.

I should know, because I cut my security teeth on the bleeding edge of PKI. In 1992, I took a position as the director of electronic commerce with a company that sought to deploy a global certificate authority (CA) that would issue the digital certificates used to process PKI. Under our plan, all other CAs would be subordinate to us, and we would sit atop a giant pyramid scheme raking in monopoly profits by charging pennies on all the billions of e-commerce transactions around the world.

The only problem was that other PKI companies were busy scheming with their own plans to take over the e-commerce world. While we were plotting against each other, we forgot to actually deploy the technology. After a few years of hand waving, PowerPoint presentations and whiteboard discussions, investors began demanding that we start earning our keep by making a profit. Silly realists!Dropping the Dot BombThe bottom soon fell out of the dotcom market, and the next thing we knew, we were all posting our résumés on Monster.com. I was lucky and found a job as CISO; others in the business were not so fortunate. Every now and again, when I have lunch with an old acquaintance, we reminisce about the good ol’ days of nonprofit technology hedonism and gossip about what company ol’ so-and-so eventually wound up with.

In retrospect, there were good reasons why PKI was joined at the hip with the dotcom boom and bust. In the early ’90s, every businessman had the same dream: a global marketplace of buyers and sellers linked together in cyberspace. The only problem was that conducting business over the Internet required authentication and encryption technologythe former to identify the buyer or the seller in a legally binding fashion, and the latter to protect the sensitive information being transmitted.

Authentication was being handled at the time (and still is) with traditional user names and passwords, which are clumsy but workable. The problem was encryption. Traditional symmetric encryption technology required that the sender and receiver both have the same encryption key. How could you get a secret encryption key to someone in cyberspace? PKI offered the solution.

Here’s how it would work. A public key could be published through a certificate issued by a trusted third party or CA. The corresponding private key could be kept under the user’s control. The sender could take the receiver’s public key from a published certificate, encrypt information using that public key and send the encrypted text to the receiver. The receiver could then decrypt the information using the private key. You could also establish the identity of people with digital signatures. I got authentication; I got encryption; I got nonrepudiation. Who could ask for anything more?

PKI originally had some great successes. All those neat little SSL sessions on the Internetyou know, the ones with the little lock at the bottom of the browser’s screenwere enabled using PKI technology. The difference is, those certificates are held on the server side. They are easily deployed because servers are administered by technical people and because companies have incentives to make their servers secure.

The problem arose with deploying certificates on the client sidewith the customers. The PKI dream was to get a certificate on every home computer. That’s where the bucks were to be had. But the technology was too complicated for Joe Consumer to understand. Not only that, but e-commerce appeared to be working perfectly well with the tried-and-tested user name and password. Before you knew it, the e-commerce market tired of all the PKI-ers’ schmaltzy talk and went about trying to make a dollar.

With the end of the dotcom era, PKI companies such as Certco and Baltimore went belly-up. VeriSign weathered the storm primarily because it was in the business of supplying the needed server-side certificates and also because it diversified. There are other PKI companies that survived but now live a zombielike existence of the undead, making just enough money to stay alive but never enough to return to their former glory days.Back to the Future?So what gives with the latest interest in PKI? Much of it stems from the fact that PKI tools have matured and are more intuitive. A second reason is that the industry is coming out of a trough in the business cycle. Third, PKI is still a viable technology that can solve certain application security problems.

This doesn’t mean we have conquered the problem of client-side authentication. If you look closely at the latest deploymentssuch as 802.1X for device authentication, digitally signed software and VPN encryptionyou’ll see that the applications use PKI for purposes other than client-side authentication.

That’s curious, especially considering the concern about phishing scams and identity theft. Citibank and American Express, for example, have launched major marketing campaigns to demonstrate how well they guard their clients from identity theft. Yet for all their talk about stopping identity theft, their customers’ only option for online banking remains user name and password. Given the threat, the time is ripe for a large-scale deployment of client-side certificates.

Instead, companies are adding security technologies that are considered more user-friendly, such as biometrics and secure ID cards, which generate a random number that’s used in addition to a password. AOL, for example, will begin offering secure IDs. Digital certificates are still seen as too difficult to deploy, administer and explain.

So what are we doing at my company? There’s no talk yet of large-scale, client-side application usage, but I have my own hidden agenda. The old saying about PKI is that the first certificate costs hundreds of thousands of dollars, and the second costs a penny. That’s because you have to build the infrastructure first. Once that is complete, you can leverage that infrastructure for a host of different applications.

Using that philosophy, I’ve managed to get my PKI project attached to an internal project with high visibility among senior managers: single sign-on. Everyone is demanding single sign-on. Once the infrastructure is built and successfully supporting that project, I can start promoting it as a cheaper security solution for other internal applicationswhich, incidentally, will include client-side applications such as S/MIME for encrypted, digitally signed e-mail.

If I’m successful internally, then I will start promoting its usage to our clients as a means of authenticating themselves for our financial service applications. Since our client base numbers in the hundreds of thousands, eventually I expect to have one of the few large-scale, consumer-based deployments of client- side certificates. Revenge will be sweet.Missed OpportunitiesBut what if the entire industry could find a way to get the last laugh? Not to sound too ’90s, but vendors are missing a great opportunity. The need for greater client-side security is there; PKI desperately needs to evolve to meet that need. One solution is placing the client certificate in firmware (for example, in chips) to make the certificate transparent to the user.

Imagine, say, an iPod with the same functionality as a PDA (messaging, calendars, cell phone and so on) and an embedded digital certificate. The device could be registered in a process that linked its preloaded certificate with the user’s account information. Then, the user could download music files using the certificate for encryption, authentication and payment. But this would be only the beginning. Other applications could be added, such as stock trading, encrypted phone conversations and online gaming.

Of course, the race goes only to the swift, and there are several competing technologies. On the low end is the secure ID, which has an enviable track record of being compact, easy to understand and easy to deploy. But the devices do nothing for encrypting sensitive information. They still require the user to input a user name and password and have to be replaced every five years because of the battery’s lifetime.

On the high end is biometrics, which seems to have become the greatest technology never actually deployed. The problem with biometrics is that, while consumers think nothing of giving their credit card to a teenage waiter, if you ask for a fingerprint, it’s “Hey pal, back off!” Biometrics are still considered much too intrusive. But, you never knowpeople’s fear of identity theft might just overcome their fear of Big Brother.

Now that we’re out of the box, let’s stretch our legs and our imaginations a bit more. Suppose we married PKI and biometrics. You could have a memory stick that contained your personal details and signing/ encryption keys. Linked to the memory stick would be a biometric fingerprint sensor. To access the personal information, you would have to pass a fingerprint scan. You could take the memory stick anywhere and plug it into whatever device needed your authentication. That sounds a lot better than toting medical records, academic transcripts, driver’s licenses and all the other flotsam and jetsam of records that accompany us throughout our lives. Sound futuristic? Not really. In fact, one company, Spyrus, already has such a device on the market.

If any of this takes off, it may be that PKI will soon make a genuine comeback. That should give me new fodder for the next time I get together with my old PKI war buddies from back in the day. I can hear it now. “Hey, did you hear that ol’ so-and-so got hired to implement a PKI solution over at….”