Hey HN!
We've just released this blog post exploring some of the changes we made over the past few years that helped us reduce memory usage by as much as 50% and improve throughput towards near-linear scalability.
We'd love to get your thoughts.
Cheers!