Hello.
Yes, I know that you never want to do premature optimization, but the monolith to write is not worth it.
There is a problem:
Writing an application like UBER. Only the web version. Expected online by 800,000 people every moment will be always online, to move, and thus to exchange data.
Plan to use:
DJANGO.
For Realtime-sharing GEODATA - Node.JS + SOCKET.IO.
Primary location: Postgres.
Storage for labels (cafes etc.): SQLlite.
Node.JS-part: horizontally scalable, new servers are automatically included in operation out of service out of service output. Each server keeps a cache with the online users. Servers interact with each other.
Question on DJANGO: what kind of architecture should I choose? Preferably with examples. Condition: horizontal scalability, working with a common RDBMS.
Also any suggestions to improve the architecture, ease of optimization, etc. in the future, coupled with constructive criticism are welcome.
Thank you all in advance. Previously with HL were not working, so, if possible, please help and examples. Sorry. :)