July 30, 2026
Queue the drama
Postgres Queues Actually Scale
Turns out the ‘Postgres can’t queue’ myth is getting roasted by its own commenters
TLDR: DBOS says a plain old database can handle huge job queues if tuned carefully, challenging the idea that you need a special message system. Commenters mostly reacted with eye-rolls, saying this is old news, while others warned the post skipped nasty real-world maintenance problems.
A fresh blog post from DBOS declared that Postgres — the database many apps already use — can power giant job queues after all, claiming it pushed 30,000 workflow runs per second across thousands of servers with the right tricks. That sounds like a big mic-drop moment… except the comment section immediately turned it into a reunion episode of “we’ve been saying this for years.” One of the loudest reactions was basically: cute post, but this isn’t shocking anymore. Commenters pointed to older systems and articles, including Rails’ default queue setup and Oban, to argue that Postgres queues scaling is less a revelation and more a repeat broadcast.
That’s where the drama kicked in. Some readers rolled their eyes at the article’s “conventional wisdom” framing, with one joking that reading Postgres queue posts feels like deja vu because people keep rediscovering the same idea. Others were less snarky and more practical: yes, Postgres can do this, but don’t skip the ugly parts. One commenter blasted the post for not addressing database “bloat,” which in plain English means old junk piling up and making the system slower and less predictable over time. And then there was the pure comedy lane: one person wondered if these posts are being published by hand or by a scheduled Claude AI task, which is exactly the kind of spicy drive-by that turns a technical thread into a spectator sport. In other words, the article said “surprise!” and the community replied, “sir, this is a rerun.”
Key Points
- •The article argues that Postgres-backed queues can scale despite a common belief that they do not.
- •It says queue workloads are difficult for Postgres because many workers polling the same table can create contention and index churn.
- •The article describes a naive FIFO dequeue approach where concurrent workers contend for the same oldest queued workflows.
- •It presents `FOR UPDATE SKIP LOCKED` as the Postgres mechanism that lets workers lock rows and skip rows already claimed by others.
- •DBOS states that its optimizations enabled 30,000 workflow executions per second across thousands of servers.