Sharded Postgres Enters the Market as a Managed Product
Neki launched as a sharded Postgres product, and the HN thread is cautiously optimistic but has sharp feedback on the launch communication. Multiple comments note the opening post doesn't explain what Neki actually is until several paragraphs in, which is a recurring HN critique of technical product launches. The questions being asked are good ones: what are the actual drawbacks besides 'new software has bugs,' what does the architecture look like, how many teams can realistically operate this in production.
The underlying interest is genuine. Sharded Postgres has historically required either extreme engineering investment or migration to a different database entirely. A managed sharding layer that preserves the Postgres interface is compelling for companies hitting scale limits on a single Postgres instance but unwilling to rewrite against a different query model.
The caution is also real. Sharding introduces distributed systems complexity, and 'new software has bugs' undersells the actual risk. Schema changes, cross-shard queries, and transaction semantics across shards are hard problems. The market for this exists, but the bar for production trust is high.
So what?
If you're approaching Postgres scale limits, Neki is worth watching but not yet betting on. The more important signal is that managed sharding is becoming a product category, which means the days of 'scale Postgres or switch databases' as your only options are ending. Evaluate carefully and demand clear answers on distributed transaction behavior before committing.