“This amounts to a ~20% decrease in disk usage, which would have been a massive boon to our disk usage issues, not to mention a significant long-term reduction in our AWS bill.”
Heap’s Postgres fleet on ZFS was choking past 80% disk. The fix was switching compression from lz4 to Zstandard and squeezing out another 20%. No new nodes and no rewrite. The millions saved were millions AWS no longer got, which is the best kind of engineering win.