We're also developing a secondary indexing module (https://github.com/RedisLabsModules/secondary) and a search module (https://github.com/RedisLabsModules/RediSearch), the direction is to integrate one of them into the JSON module.

Wise words -- a good set of indices are the first thing you need for real problems with realistic data sizes.

Right now the problem that ReJSON tries to solve is very specific and typical for Redis users: You save JSON blobs on Redis (usually session data), now you want to get just a part of a JSON key, or manipulate just one element in a specific key.

Until now you couldn't do it without loading the entire object into memory. ReJSON just allows you to manipulate that object or retrieve just a part of it efficiently. For this use case we do not need and will never need indexing.

By that logic, redis in general would not be useful for "real problems", whatever that is.