At ~35TB Postgres we ended up separating pg_wal onto its own volume first — write amplification there was quietly eating 20–30% of our NVMe throughput before we caught it.
We’re on Hetzner too, using multiple 10TB volumes joined with ZFS. Total size right now is around 45TB, mostly cold data. For hot tables we keep them on NVMe (~20TB). It’s not perfect, but works fine so far.
At ~35TB Postgres we ended up separating pg_wal onto its own volume first — write amplification there was quietly eating 20–30% of our NVMe throughput before we caught it.
Please tell me you don't store (binary) BLOBS in your db.
We’re on Hetzner too, using multiple 10TB volumes joined with ZFS. Total size right now is around 45TB, mostly cold data. For hot tables we keep them on NVMe (~20TB). It’s not perfect, but works fine so far.