How do you evaluate the trade-offs between symmetric and asymmetric encryption in different scenarios?

Aside of the challenge requiring a secure out of band communication to transfer the key allowing both encryption and decryption, there is no way for both sender and recipient to know they are the only ones accessing the encrypted content. This technique, while the key is hijacked, allow man in the middle attack, as the integrity can't be verified. Interception could occur, content decrypted and re encrypted just by guessing or obtaining the unique shared secret. This can lead to a fake sense of security.

One of the great advantages of asymmetric encryption is that it does bring the ability to establish non repudiation of shared content, assuming the private key is kept confidential. Ideally, in such implementation, the private key should also be password protected to avoid potential identity usurpation.

The TLS (Transport Layer Security) protocol, which secures HTTPS communication, uses asymmetric encryption (like RSA or ECC) to exchange a symmetric key. Once exchanged, symmetric encryption (like AES) is used for actual data transmission, ensuring both speed and security. In a recent engagement with a healthcare platform, we implemented hybrid encryption to protect sensitive patient data during transmission between clients and the server. The combination allowed us to leverage the scalability and security of asymmetric encryption while ensuring high performance when handling large volumes of data using symmetric encryption.

In layman's terms quantum computers can bring the ability to test any key combination at once, which makes any mathematical based encryption (such as encryption based on prime number factorization) are therefore rendered useless as they can instantly being solved. One of the key issues as we know these will be capabilities of quantum computers, this is not yet practically exploited. The issue that is brought in this timely issue is known as "record now, decrypt later", which means saving current encrypted data flows, aiming a replaying them and decrypting them once quantum computer is successfully programmed to do so. It is way more critical than it seems as any sensitive information exchanged today, might become public tomorrow.

While Post-Quantum encryption (or PQE) is widely discussed, it remains very theoretical given we do not yet completely understand the capabilities coming with quantum computers. Another consideration is that these PQE implementation are designed recently and at early stages. This means that aside of the hope for quantum proof encryption capabilities, design flaw may also lead to inefficiency. We saw in 2022 that PQE encryption algorithm got broken by a traditional computer system. This means that reaching maturity in this field will take time. Ideally, very sensitive data should no longer transit via public network (Internet, or public cloud) because we can't protect information on this support for a fairly long future.

The present threat being "record now, decrypt later", a thoughtful risk analysis must be run over any information using public transit networks (internet) or cloud computing, as we can't protect information as of today against tomorrow threat. We also need to assess historical data, given the amount of BGP traffic hijack in the past years, meaning that all previously exchanged information is potentially at risk. This is a tremendous inventory task, and the only way to render these decryption useless is to render the data irrelevant. An heavy work is requested, making sure decade old transmissions are reviewed, and developing quantum proof encryption. Existing industrial systems must be considered too (satellites, infrastructure etc).