Is it possible to aproximate or that someone tell us his experience about the order of magnitude of users for a container in Galaxy? We know that this depends on each app but maybe it’s possible for someone to tell us his numbers.
Our container now is simplest one , but if someone has information for any other type of container it’s welcome too. By now our is 512RAM and 0.5 ECU.
Our complex app (without microservices by now) uses 130MB aprox for a single user connected (Galaxy and Kadira show similar data) and 0.03 ECU.
The amount of users you’ll be able to serve completely depends on the app, you could have an app that is only able to serve 1 user per instance or you could have an app that is able to server 1000 users per instance. Most apps will be somewhere between (more realistically you’re probably speaking 100-500 users per instance)
The only way you can determine this in detail is by testing. You could start with doing it manually, have your friends browse the app etc. The next step would be to do real stress testing, you could use either something like meteor-down (https://github.com/meteorhacks/meteor-down) or some cloud service like https://www.blitz.io/.
(more realistically you’re probably speaking 100-500 users per instance)