CertiGuide to Security+  9  Chapter 4:  Basics of Cryptography (Domain 4.0; 15%)       9  4.5  Key Management and Certificate Lifecycles            9  4.5.10  Key Usage

4.5.10  Key Usage 1 4.6  Summary

Conventional hierarchical PKI uses a single key pair – one public key and one private key. According to VeriSign, “Key Pairs are used for one or more of three basic purposes: encryption, authentication and non-repudiation.” A single key used for multiple purposes violates non-repudiation⁴¹⁰. Up until now, we’ve primarily been discussing situations in which each individual has only one key pair.

RSA recommends that you assign two key pairs per person -- one key pair for signing messages, providing authentication and non-repudiation, and a second key pair for encryption. This allows someone to recover the encryption key and decrypt documents that were encrypted using it, without their gaining the ability to sign documents with that user’s private key as well (which could lead to forgery and violation of non-repudiation).⁴¹¹

With a single key pair, this protection is not available. When a single key pair is used, key recovery gives the individual recovering the key the ability to masquerade as the user whose key was recovered, if desired, in addition to allowing them to encrypt/decrypt communications from and to that user.

Since an organization already has to have infrastructure in place to manage a single set of keys, it’s generally not that much more difficult, from an administrative point of view, to add a second set of keys for each user (as long as any key management software/hardware used supports this). Because of the low incremental cost and the potential value of implementing a PKI with dual key pairs, it’s something to consider.

Note that when using dual key pairs, you may wish to treat the archiving and escrow of signing keys different than encryption keys. The reason for this is that if a signing private key is lost, it’s no big deal – just generate another key pair for future messages (the public half of the old signing key will still be “good” for reading messages signed with the old private key). Additionally, if the employee were no longer with the company, although you might have a need to go back and decrypt messages encrypted with his or her private key, you wouldn’t need to digitally sign a message with his or her identity. Contrast this with the loss of an encryption private key – which means that any outstanding messages to the recipient, created with the recipient’s public key, cannot be read⁴¹².

4.5.10  Key Usage 1 4.6  Summary