he Data Encryption Standard, or DES, was developed in the 1970s as a cooperative effort between IBM, the National Security Agency and some other governmental bodies.

It was a first try at producing a high-quality, government-approved cryptographic algorithm for public use. They did a really good job.

Several cryptographers have spent the majority of their careers searching for some secret hole in DES, some tricky way of getting through a back door that they believed was there. But as it turns out, no such vulnerabilities were found to exist, and over time DES became accepted as the commercial cryptographic algorithm of choice.

As time moved on, financial institutions started using DES and began to create security infrastructures to protect their growing number of electronic transactions. However, most folks really didn't take the whole security thing too seriously.

It was not unusual to have thousands of ATMs keyed with the same single DES key that was never changed. And just so I don't pick solely on the financial industry, the satellite and cable TV industries did similar things. There were probably hundreds of thousands of set-top boxes in the 1970s and 1980s that had the same single DES key permanently installed so that any technician could read it.

The Technology Revolution

Then, several things began to happen. Computers got faster, crooks got smarter (or at least if not smarter, they started applying themselves) and computer crime started happening regularly.

Note that many of the events called computer crimes were, and still are, really social engineering exploits expedited by computer. That list includes everything from Nigerian bank account scams moving from paper mail to faster and less expensive e-mail to collusion between financial workers routing wire transfers to destinations other than those the account holders had requested.

However, real computer crimes were occurring, too. The bad guys started figuring out ways to penetrate computers at financial institutions (and other places), and for a while it was the digital Old West. For a time, no one bothered trying to crack DES, or any other algorithm, because it was just too hard; there were so many easier ways to successfully attack organizations and steal their data.

So in response, basic perimeter security was vastly improved. Many organizations did a good enough job to prevent most intrusions, but not all.

There have been several well-reported incidents in the news where intruders have stolen collections of personal and financial information for ransom, or for use in identity theft. But for the most part, good practice can prevent external intrusion.

In addition, protection of "data-at-rest" is the new buzz, and products are being introduced (or in several cases just being noticed) to protect data and databases.

The biggest problems remaining are internal issues. Insiders account for about 70% of all computer crime. Again, good practice can eliminate most of the threat as well, but we'll leave that discussion for another time.

Back to DES. Now that we had plugged most of the easy holes, cracking DES started looking like a good idea again. Over the intervening time, computers had gotten a lot faster and less expensive, and the Internet had become a freely distributed computing engine.

Cracking the Code

In the late 1990s, the Electronic Frontier Foundation (EFF) built a machine called Deep Crack with about 250,000 DES engines that could all run simultaneously. Each DES engine would brute-force-search a portion of the 56-bit DES key space to find the key (a DES key has 64 bits, but eight bits are used for parity, so only 56 bits are actually used as the key, so the key space is two to the 56th power).

It worked. Amazingly enough, an organization supported largely by donations, without the resources of any large corporation or government, built the first cryptographic doomsday machine.

Then they got even smarter. They teamed up with Distributed.net and set up a program whereby anyone with an Internet-connected computer could volunteer a portion of their computer's resources (usually when the screen saver was running and the user was not actually working on the computer) to search part of the DES key space.

The result was that in February 2000, the team of the EFF and Distributed.net cracked a DES key for the RSA Challenge in about 22-and-a-half hours. That was the day that DES died.

In some sense, this was a good thing. It was a real wake-up call and started a change that we are still working on today. First, the industry needed a new algorithm; second, the policies and procedures that had been in place since the beginning of electronic commerce needed to be reviewed and improved.

For the new algorithm, choices needed to be made. The new algorithm had to be secure. In fact, it had to be secure enough to have a reasonably long lifetime, as no one wanted to do this again anytime soon, and with good reason: it's expensive, time consuming and complicated.

There were two major candidates for replacing DES: Triple DES (sometimes called TDES or 3DES), or Advanced Encryption Standard (AES). Based on tried-and-true DES, 3DES uses the original DES algorithm three times to encrypt the data. Using either two or three 56-bit DES keys, 3DES is strong enough to see us through the next generation of security. Most experts think that this will be for at least 15 - 20 years, but a breakthrough in computing could always change that.

The other choice for the new algorithm was AES. At the time when the financial industry needed the new algorithm, the National Institute of Standards and Technology (NIST) was in the process of running a public showdown to pick its replacement for DES. The NIST competition would bring a new algorithm to the table, one that would be secure, computationally efficient and have a long projected lifetime. The competition was a great thing Algorithms were submitted from all over the world. All of them were subjected to public and governmental review, and all were tested for performance and efficiency.

As it turned out, the choice for AES was a good one. The algorithm chosen, Rijndael, developed by a Swedish team, is more computationally efficient than 3DES by a wide margin and supports key lengths up to 256 bits. It will be here for a long time.

However, the industry was not prepared to wait or experiment. We already had 3DES and most everyone was familiar with it. So that's where we went. The migration was on.

Encryption and the Future

Over the last several years, as we all know, the big issue in encryption for the financial industry has been the change from single DES to Triple DES.

And we are still on the way, partly because making all of those changes is a lot of work, and in many cases, requires upgrading hardware, software or both. But there is still the "other" part of the job. The cryptographic algorithm used is only part of the equation.

We also need to update all of the security policies and practices that revolve around the use of the cryptographic algorithm. That is true for any algorithm, whether it be DES, 3DES, AES or something that we haven't imagined yet. As of now, poor practices such as using one key for many ATMs, POS devices, etc., need to stop.

New rules and standards already require that each device have its own key. Likewise, it is now, or soon will be, a requirement to change keys at regular intervals. As evidenced by the recent theft of information collections, new policies and practices must be developed and correctly used to protect the rest of our data as well.

Data-at-rest and database encryption are coming into use; and the protocols that we use to secure our communications, from SSL for Web sites to the VPN we use to connect to the office from home or another site, all need to be evaluated and the security verified.

In the future, the need for data security and integrity will continue to require our careful thought and consideration. The problems will not go away; rather, they will evolve, and the bar will continue to rise. The bad guys will get better, and because of that, so will we.

We will likely move on to AES someday, maybe not for five, seven or 10 years, but we will. We will also make more and better use of public key cryptography (RSA and ECC) to distribute keys and data, for time stamping and integrity verification, and for non-repudiation (which is another new use for these tools that is becoming common).

It is also likely that all of our data will be encrypted whenever it is stored. Most if not all of our networks will be encrypted or behind firewalls and security appliances that are much more sophisticated and secure than what we have now. There is a lot going on in the application of cryptography to support our industry. It started with DES and is moving on to meet the new issues.

Link to original article: www.atmmarketplace.com/research.htm?article_id=23765&pavilion=126&step=story

Steve Weingart is the Chief Technology Officer at Bulverde, Texas-based Futurex, a company that produces cryptographic hardware. He has worked in cryptography for more than 20 years.