Describe assumptions and guidelines on key management

[tQsW]

No description provided.

[tQsW]

This draft could address the key discovery mechanism, and leave the process to provision keys to implementations. Specifically, identifier or name based discovery could be a good candidate to accommodate different scales, and is also aligned with the current WebCrypto spec on the key discovery.

A small managed network may share a common and small set of keys pushed by an admin periodically via a secure channel. When stored as key-value pairs, where a name is mapped to a set of encryption keys, the set of encryption keys can retrieved with the given name or id if such a mapping exists, and trial decryption can be used if the set contains more than one key. A fail-to-decrypt error can be declared when no encryption key set is present, the set is empty or the decryption fails with the retrieved key set.

In a more advanced form, the encryption key store can be servicified, not necessarily co-located on the same device, and use appropriate authentication techniques. This approach can be extended to support flexible key rotation and also origin-specific keys if needed.

[juberti]

Do you have some text in mind here?