Backend is not horizontally scalable or durable

[brandon-otoole]

The room data and drawing histories are all stored in memory within the backend. Server crashes, restarts will clear all history.

For realtime speed, something like redis should be used. To enable horizontal scalability, a message queue like redis pub/sub, or other needs to be implemented between nodes For durability, either Redis backup, or a typical database should be used

[brandon-otoole]

The major components needed for durability of data:

• Design the redis data definition needed to store and manage histories
• implement a data interface layer to read/write the histories
• initially deploy redis as a container enables durability but not scalability

Components needed for scalability:

• some form of message queue or other server to server communication
• eventually a more robust deployment system like k3s, Kubernetes, etc

[brandon-otoole]

Given that durability and scalability are different concerns. I will first focus on durability of data by deploying reddis as a container. This works well with the current deployment strategy using docker compose.