August 3, 2026
Backup bragging meets comment chaos
Massively Parallel Postgres Backups
PlanetScale says backups can fly at insane speed, but commenters want receipts
TLDR: PlanetScale says it can save massive databases quickly by using temporary cloud machines so the live app barely feels it. Commenters split between admiring the clever scaling trick and side-eyeing the company for missing other basics, turning a backup post into a mini credibility debate.
PlanetScale just pulled back the curtain on how it saves giant databases without slowing down the apps people are using. In plain English: instead of stressing the live system, it spins up a bunch of temporary cloud computers, rebuilds the last saved copy, then fast-forwards all the recent changes until everything matches the present. The big flex is scale — we’re talking enormous data piles copied in hours, not days — and the company frames it as the kind of invisible magic customers only notice when disaster strikes.
But the real action is in the replies. The author jumped in to say the deeper lesson isn’t just "look how fast our backups are," but how to make an ordinary tool scale up to absurd size by stacking good existing parts together. That got approving nods from the practical crowd who love a solid engineering philosophy. Then came the classic comment-section record scratch: one user basically said, cool story, but where are the Postgres version upgrades? Ouch. Suddenly the vibe shifted from victory lap to accountability check.
That tension is the whole soap opera here: supporters see smart, boring infrastructure done right — the kind of thing that quietly saves companies from catastrophe. Skeptics hear a glossy performance and immediately ask what’s still missing. It’s a familiar internet plotline: one side posts the giant machine, the other side points at the loose screw. And honestly? That contrast is what makes this backup drama so deliciously watchable.
Key Points
- •PlanetScale describes a backup system that creates consistent, encrypted snapshots of busy databases every 12 hours while avoiding impact on production queries.
- •For sharded Postgres databases, the system relies on standard Postgres backups as building blocks and uses Neki to coordinate backups across multiple shards.
- •The backup method combines filesystem backups with archived WAL replay so that changes made during the backup window are reapplied to produce a consistent result.
- •Instead of backing up directly from production primaries or serving replicas, PlanetScale launches temporary cloud instances, one per shard, to perform most of the backup work.
- •PlanetScale says the approach can back up petabyte-scale databases in hours with throughput above 50 GB/s, while also validating that previous backups can be restored and replayed.