Good work! Caching is one of the 2 hardest problems in Computer science. Not because building a cache is hard, but cache invalidation/replacement is complex when you consider the system as a whole. Integrating the cache and preventing user from seeing the stale value is not a trivial task (if you want to support highly concurrent access). MySQL removed query caching from 8.0 because of the mess that query caches are. Enforcing transaction isolation levels in a database with caching is non-trivial.
Exactly. Doing caching in a transactional MVCC system is hard. We plan to do careful data caching in Dgraph with this. Block level cache in Badger would be easy. So would be key with version level cache in Badger.
