# DigitalOcean, this is not how you launch an "Advanced" Postgres database... — retained record

> Retained reddit record from reddit:database.

Canonical: https://highsignal.app/data/records/c81f8d97876cfb08

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:database)  published 2026-11-19 16:10 ingested 2026-11-20 02:39

[original source ↗](https://www.reddit.com/r/Database/comments/1y0d8sg/digitalocean_this_is_not_how_you_launch_an/)

## retained text



```
Hey guys, for those of you using DigitalOcean: you may or may not know they're phasing out the Standard-tier Postgres plans in favour of a new Advanced tier on November 30.

Naturally, I benchmarked both to see if there was a difference. The results were so atrocious they defied my belief.

\- Up to **436% higher p99 write latency** (6.7 ms → 35.9 ms).

\- **\~50%-70% fewer** **write and read throughput**

the new tier tops out at \~54k reads/s & 6.9k writes/s, versus the old Standard tier's 204k reads/s & 26.5k writes/s max.

\- 5x slower startup times: 3m 19s → 16m 06s, averaged across 15 regions.

The results were averaged over 3 runs per DB, for all possible DB sizes (2->32vCPU).

This was done over two weeks of testing, running around 60 hours of wall-clock time and going through 100+ databases. It cost me a few hundred dollars to run.

If interested, [I posted the full results & methodology on my blog post.](https://smokentar.com/blog/digitalocean-postgres-benchmark/)

I believe the bottleneck is a fixed-process PgBouncer deployment. But I don't consider myself an expert in databases so would really appreciate if you could double-check my logic.
```



document fetched 2026-11-20 00:39 · canonical https://reddit.com/r/Database/comments/1y0d8sg/digitalocean_this_is_not_how_you_launch_an
