Judoscale on Northflank: Powerful Autoscaling for Hosted and BYOC Deployments

We’re continuing to bring Judoscale to the cloud hosting platforms we love. Today that platform is Northflank.

Northflank is interesting because you get two deployment models under one roof: a fully hosted option (like Heroku, Render, etc.), and bring-your-own-cloud (BYOC) on your own AWS or GCP account—for pricing, compliance, or whatever reason you have for keeping the infra in your cloud. The important part, at least for me, is that both options share the same interface and the same user experience. I’ve been deploying demo apps on the fully hosted side, and our early-access customers are using Judoscale against the BYOC approach. From Judoscale’s point of view (and yours), both work exactly the same.

Do you even need Judoscale on Northflank? I’ll admit Northflank’s built-in autoscaler is no joke. You can scale on CPU, memory, and RPS, and if that’s not enough you can emit custom Prometheus metrics and autoscale on just about anything. It’s very good for a built-in autoscaler. So when should you reach for Judoscale? When you want a bit more control over how the scaling behaves in addition to what metric you watch. Judoscale lets you dial in how aggressively you scale up, how patiently you scale back down, and which signals you actually care about: request queue time for web traffic, job queue latency for Sidekiq, Celery, BullMQ, and friends.

We recommend starting with Northflank’s built-in autoscaler, and upgrading to Judoscale when you want more control or visibility into your autoscaling.

If you’re on Northflank and want that kind of control, create a free Judoscale account, connect your project, and you’ll be guided through setup in just a few minutes. Learn more in our Northflank integration docs.