Key Lengths and Lifetimes

Before discussing the lifecycle of cryptographic keys, we need to consider a couple of the properties of the keys themselves, most significantly the key length.

We already know that longer keys are better from a security perspective in general (but certainly not by default). Longer symmetric keys take more time to search for exhaustively, and longer public-key pairs tend to make the underlying computational problem on which a public-key cryptosystem is based harder to solve. So there is certainly a case for making keys as long as possible.

However, a cryptographic computation normally takes longer if the key is longer. In addition, longer keys involve greater storage and distribution overheads, which is why longer keys are less efficient in several important respects. Thus, key length tends to be based on an efficiency/security trade-off. We normally want keys to be ‘long enough’ but not longer than this.

Key lifetimes

The key length issue is closely linked to the intended lifetime (also often referred to as the cryptoperiod) of a cryptographic key. By this, we mean the key can only be used for a specified period, during which it’s regarded as being live. Once this lifetime has been exceeded, the key is regarded as expired and should no longer be used. At this point, it may need to be archived or perhaps destroyed.