There are two method to using hubs, hubs cloud and Full Self-Hosted Hubs (this repo's about)
Correct me if im wrong, As i know the different are
- Only in AWS or Digital Ocean
- Partialy open source code (don't have access to reticulum code) only hubs and spoke code
- Email service easy to setup (SMTP)
- Require less technical skill
- I don't know how to backup database or storage on this method
- Hosting anywhere
- Full access all code
- You can fully optimize the code, storage, Directly fix all bugs
- You must setting up SMTP for email verification (i don't do this i just editing reticulum code to match my needs)
- Require more technical skill (System architecture skill, dev ops and the program it self)
How many maximal user can interact? you must doing simulation research.
I will share little bit of my research.
Let's define the parameter what makes server can serve hubs user interaction at the same time (concurency)
This is the estimation.
| Action | Memory (MB) |
|---|---|
| 27 Online together in one room. not moving | 80 |
| 1 Share Screen | 50 |
| 1 Moving user | 30 |
| 1 User Not Moving | 3 |
Assume one room is 25 user multiple by 10 if everybody is moving continuously is cost 250 MB
The hubs require less cpu usage if the user just a few. I haven't test it with a lot of user
There are two kind of bandwith, Bandwith international (overseas) and local bandwith.
So the user will be able to access a server that is closer. thus making the access speed faster
Making web bot that can simulate user interaction on the room. (walking, do click, etc).
Also the bot can replicate him self. so we can simulate how many user can access and interact at the same time.