Autoscaling on a Sovereign Cloud: Judoscale for Scalingo

Judoscale now supports Scalingo. If you’ve used Heroku, Scalingo will feel familiar: push your code, run web and worker containers, and let the platform handle the infrastructure underneath. Scalingo is a sovereign European cloud: it hosts apps in Paris datacenters and keeps your infrastructure under European jurisdiction. For public-sector and regulated workloads, its SecNumCloud-backed region goes further by running on French infrastructure that meets France’s strict SecNumCloud security standard.

We built this integration because several customers migrated from Heroku to Scalingo and wanted to bring queue-time autoscaling with them. Yespark was one of those customers:

Judoscale has been live in our Scalingo app for several weeks now with really solid results. Autoscaling is doing its job.

— Navid Emad, Full-Stack Developer at Yespark

This integration works just like our other platform integrations. Judoscale reads request queue time from web containers and job queue time from workers, then changes the container count through Scalingo’s API. Scalingo’s native autoscaler can use traffic and response metrics for web containers, but non-web containers are limited to CPU, RAM, and swap. I/O-heavy jobs can back up your job queues while CPU stays flat, so Judoscale scales workers based on job queues instead.

Get started by creating a free Judoscale account and connecting your Scalingo app using an API token. We support Ruby, Python, Node.js, and Java apps, along with their common background job systems. See our Scalingo setup docs for the full walkthrough.