Pinned
PlanetScale
@PlanetScale
The fastest and most reliable database for Postgres & Vitess
Discord: pscale.link/community
Support: @planetscalehelp
planetscalestatus.com
Joined February 2018
- “Vitess for Postgres” is a useful shorthand for the ambition behind Neki, but it understates the architecture we have built. At the core is the data topology, the configuration that maps logical Postgres tables to shards and routes queries between them.
- We have two more awesome lightning talks lined up for the SF Database Meetup on August 19th: - How we rewrote Postgres in Rust and made it 300x faster for analytics by @mmalisper - The OLTP of Streaming by @shikhrr Grab a seat:
- A Postgres transaction opens 65 subtransactions. The result? 7,200 TPS → 160 TPS performance drop. We dive deep into Postgres internals in our latest blog, showing why this happens and the dangers of using too many subtransactions in Postgres.
- Reduce parallelism. Increase throughput. Sounds backwards, but past a certain point, each additional in-flight query can make all the others slower. We recently observed this in a customer's Vitess database. How it happened, and the fix:
