Linuxcontainers.org says Incus “offers a seamless cloud-like experience scaling from a developer’s laptop to a full cluster of up to 50 servers.”
At my current scale, 50 servers is probably enough. But I’m just wondering if anyone is running around that 50 server scale or larger, and if so, are there problems that emerge at that larger scale (cluster state sync, storage migrations, etc)?
Thanks for taking a look!
We’ve done scaling up to 250 servers in a cluster at this point, there is no real issue with scaling that up, total number of ongoing operations that need database writes is the biggest bottleneck, but running instances doesn’t really impact that.
The reason for the 50 is blast radius. We prefer folks run multiple standalone clusters than a single very large cluster as if something goes terribly wrong with a database update, you could find yourself with the control plane of the entire environment going down.
That makes a lot of sense. Also, thanks for the warning re: blast radius.