Post

PlanetScale now supports Postgres.

Your performance comparisons are very skewed to favour you. Not exactly screaming credibility here…

We cover all the details of how they were conducted on the blog and benchmark pages, and welcome input. What about it seems skewed to you?

Thanks for not ignoring me! As a Crunchy customer, I was excited to read your comparison post. But two things stood out that make the comparison seem unfair.

Also, Crunchy doesn't seem configured with pgbouncer while PS seems to have a pooler by default. That alone makes this comparison sketchy. Would love to see the blog post updated with that option enabled. Don't want to blindly hate here - happy to be proven wrong! ;)

At the end of the day, I do not expect you guys to do too much magic here compared to a managed Postgres at Crunchy. Also, I do not deny that local NVMEs make a performance difference, I just want to see actual numbers on what specifically they do.

You're right about replication. Due to the nature of direct attached nvmes in the cloud we need to maintain multiple copies for resiliency. But these copies are also used to for HA and can take some read load from the primary, so they serve multiple purposes.