How Much Crypto You Really Need

Last month an international team of researchers announced they had managed to factor a 768-bit RSA key. This raises interesting questions about handling encryption and planning ahead in your security strategy.I am not a cryptographer and therefore not an authority on the RSA algorithm. However, I would like to discuss the strategic use of encryption and how that impacts our decision on choosing and implementing an encryption scheme.

In the introduction to their paper (PDF), the researchers wrote:

"Because the first factorization of a 512-bit RSA modulus was reported only a decade ago (cf. [7]) it is not unreasonable to expect that 1024-bit RSA moduli can be factored well within the next decade by an academic effort such as ours or the one in [7]. Thus, it would be prudent to phase out usage of 1024-bit RSA within the next three to four years." The researchers make it clear that this is not an immediate threat, and while the use of larger keys should be phased in, there are no immediate security ramifications.

When choosing an encryption algorithm, you must first ask yourself what your purpose is -- in other words, conduct a risk analysis. What do you want to protect? How important is it? How vulnerable is it? What is the threat? Who is the threat?

If your biggest risk is your neighbor, then you can feel relatively secure using off-the-shelf encryption without worrying about too many details.

However, even if you are not a nation-state with an opponent that will invest unlimited resources to get your information, there are three main questions you should ask:

1. How long into the future would I like this information to remain secret? Based on this answer you can consult with an industry analyst as to projected computer power changes in the coming years. Any foreseeable breakthroughs in the math that could reduce the time needed to break the encryption. Then based on two factors, make the call: How paranoid do you feel you need to be according to your risk assessment, judged against the functionality you require and implementation costs?

2. Does my opponent have the resources to deal with this encryption? To break modern encryption on a PC could take longer than the life of the universe. Don't be confused by this statement: Consider what else your opponent might be willing to do to get your information. Encryption makes us feel safer, but it does not equate security.

As an alternative, also consider that encryption is a secret, and you might want to use several encryption schemes so as to not make one too secure to work with; you won't be able to trust people who run your daily operations to use it.

3. Because the algorithm is rarely the weakest link in real-world attacks, have I taken care of the implementation? Most attacks against encryption systems are against the implementation rather than the algorithm -- be it the programming, which should be done and reviewed by experts, or the procedures by which the encryption is put to use.

Further, side-channel attacks ranging from the less likely, such as TEMPEST (electromagnetic emanations) to Trojan horses (which are a significant threat), and from stolen laptops to buying off an employee, must not be left to chance. These should all be considered and planned for. The algorithm alone does not make you safe.

Follow Gadi Evron on Twitter: http://twitter.com/gadievron.

Gadi Evron is an independent security strategist based in Israel. Special to Dark Reading.