Offload processing of data away from the endpoints.

[RealistikDash]

Currently, all of the processing for user data is done during handling the endpoint. This is bad from a scaling standpoint. Ideally, the bare minimum should be done during the request itself (such as just a simple DB insert).

This should be off-loaded to processors running ideally on a separate process. These will be distributed over AMQP and handled. Likely will require an additional field per model dedicated to keeping track of whether these have already been processed and can be displayed to other users.

Some of the places I can see this being implemented are:

Levels:

• Meili indexing
• Level content analysis

Users:

• Meili indexing
• Stats anticheat
• Stats analysis

Messages:

• Spam filter

User comments:

• Spam filter