Good read. Reminds me of the 1 billion row aggregation challenge, especially the perfect hashing part, all the top entries all used it.

https://github.com/gunnarmorling/1brc