I personally tend to put all my data for analysis into clickhouse (which is written in C++ FWIW). My side project actually is all about ingesting JSON data into clickhouse and making it super easy to query, visualize, monitor and alert on JSON data.
You certainly pay a computation cost everytime you want to scan the data, but I think the flexibility more than makes up for it. You can take a read about how and why I built GraphJSON here if you're interested https://www.graphjson.com/guides/about
ps - we're built on Clickhouse at Luabase (https://luabase.com/) and hiring across all role, send me a note (mike@luabase.com) if you're into web3+analytics.
You certainly pay a computation cost everytime you want to scan the data, but I think the flexibility more than makes up for it. You can take a read about how and why I built GraphJSON here if you're interested https://www.graphjson.com/guides/about