Breaking down our project

We are building a simulation platform where users can log in and interact with fake simulated people via chatbots to see what they say to the fake people. We want to see what the users say to the fake people at the end of the simulation session. The problem is that there are a lot of messages being sent to a fake person in a short amount of time. If we hit the database every time a message is sent, then we will put our database under a lot of strain. Let us say that a user asks a question every 20 seconds; this means we will be hitting the database 6 times per minute as there is a question and an answer per interaction. If we have 800 users running a session at the same time, then we can have up to 4,800 hits a minute. This can put a strain on the database. To reduce the strain on the database, we can build a server in Hyper that caches the chats and periodically sends multiple questions and answers to the database. Before moving forward, this is a chance...