-
Sorry, I could only post in Off-Topic...
I was curious how Cloudron handles scaling. I was considering using Cloudron for about 50,000-100,000 users, maximum of 2,000-5,000 online at a time. 300 instances of a wiki, 300 forums, (300?) synapse servers, and 300 SOGo, Radicale, Easy!Appointments, or some other calendaring software, and one primary (large instance for everyone) Element Client, forum, wiki, and calendaring. This is for federating 300 organizations.
How well could Cloudron actually handle all of that? What approach would be optimal? Could Cloudron use managed databases for instance to scale better? What sort of VPS size would I likely need at that point?
-
G girish moved this topic from Off-topic on
-
I don't have any empirical numbers, but I am doubtful that Cloudron would be a good fit for you (at least at this point in time).
So far Cloudron is a single server system, not able to make use of multiple nodes for hosting the actual functionality. There is some work going on in this direction, but so far this looks more like a "install the app on server x instead of locally" type of multiple node setup.
50.000 to 100.000 users is also a serious number of entries in the user list, which will not only make Cloudron choke, but also any other application that is trying to list them that has not specifically been programmed to be able to handle these amount of users (I am not even talking this amount of user actually using the system, but even just a "give me a list of users" in e.g. the wiki software you want to use).
If you are serious with your request then I would recommend to reach out to the Cloudron developers directly with information about the type of servers you have available and the size of the budget you have set aside for the software side to see if it makes sense to do specialised custom development for your use case.
-
My experience with Cloudron tells me that scaling for a larger user base only works via a "larger" VM. There is no automatic scaling or scaling of database clusters and from my understanding of the roadmap this kind of scaling is not planned for the next versions.
So if you have buzzwords like Kubernetes or Hyperscaler in your head... that's not the purpose of Cloudron.
But I have heard of Cloudron administrators managing up to 3K users on a single instance. That should not be the limiting factor
Handling a Wiki with 3k users?! I think that shouldn't a problem. Most of the time users read.
Synapse for up to 3k users? IMHO possible.
SoGO for up to 3k users? Sounds like a lot of RAM for IMAP sessions.
Using Nextcloud on a single instance Cloudron for up to 3K user? Forget it
IMHO the only strategy could be: one instance per organisation. Start with one organisation on a let's say 64 RAM up to 32 CPU instance and find out, how the people interact simultaniously with the apps. Install on a separate small vps a monitoring system and get some metrics. If you find out, that you need more RAM/CPU, migrate the whole Cloudron instance to a bigger one. It's easy to migrate and should be done in less than an hour.
If it's a good fit, migrate other organisations to "their" dedicated Cloudron instances. Cross your fingers that the upcoming Cloudron Multihost feature will cover some of your use cases. You have the best chance if you join the discussion.
Hope this helps.
-
While looking at something different in the documentation i came across the following statement in https://docs.cloudron.io/packaging/cheat-sheet/#memory-limit
Design your application runtime for concurrent use by 100s of users. The Cloudron is not designed for concurrent access by 1000s of users
