Native, full-stack autoscaling on Upsun: PostgreSQL and MariaDB read replicas now autoscale

Application autoscaling has existed for years. Database autoscaling has lagged behind. Today, that changes on Upsun.

Managed PostgreSQL and MariaDB read replicas now autoscale, joining worker autoscaling in production since March 2026. Apps, workers, and managed databases all autoscalead replicas saturate first, connection pools max, someone getse on one platform - natively, without a third-party database engine or a Kubernetes layer to operate.

Three operational headaches just disappeared. Here's what changed.

The 2am page about your database, gone

The application usually survives the traffic spike. The database is where things break. Read replicas saturate first, queries queue, someone gets paged.

Starting today, Upsun automatically adjusts read replica capacity in response to CPU and memory usage. When workload pressure rises on the replicas, more are added; when pressure falls, capacity is released. Scaling is dynamic - no restart, no failover, no plan-upgrade window.

The outcome: traffic spikes from launches, campaigns, or seasonal events stop turning into database incidents.

The pre-scale ritual, retired

The calendar reminder before a high-traffic event to "manually bump worker capacity" - and the follow-up reminder to "manually scale back down two weeks later" that nobody actioned - those go away.

Worker autoscaling has been live since March 2026, scaling on CPU and memory usage. Payment processing, video transcoding, AI inference, scheduled batch - anything async - stopped backing up during traffic spikes. Today's release extends the same primitive to managed databases.

The outcome: the DevOps tax of manually configuring scaling tools goes away. Routine capacity events stop being calendar items.

The peak-capacity bill, normalized

Most teams over-provision the database tier - sizing it for the worst peak hour of the worst peak day, paying for that capacity 24 hours a day. For cyclical workloads, that's a significant share of capacity sitting unused most of the time, showing up on every invoice.

Upsun Autoscaling scales down as well as up. Capacity is removed when traffic falls.

On the invoice: spend tracks actual usage.

How it actually works

Upsun monitors CPU and memory usage across worker pools and managed databases - the signals that directly reflect resource pressure. When workloads push CPU or memory up, Upsun adds capacity; when the load falls, capacity is released. Replicas join and leave the running cluster without restart or failover.

Native, full-stack autoscaling

Autoscaling shouldn't stop at the application layer. With database read replica autoscaling, Upsun now scales applications, workers, and managed databases together - natively, on one platform. No third-party database engine to integrate. No Kubernetes layer to operate. Infrastructure that scales as a system, not as disconnected parts.

• Ready to get started? Enable autoscaling on your project
• Have questions? Talk to us