Cleaner data storage architecture

[dsonck92]

Currently, the service layer of the API has a lot of contamination within its business logic. The proposal is to change the software layers to the following:

• Routers, talk to the service layer and are responsible for changing the "business logic" models into "API" models
• Services, contain core business logic like:
  • Validating the action can reasonably be performed
  • Validating the action is allowed to be performed
  • Starting all actions in the way they are supposed to
• Data abstraction layer, interfaces defining the core concepts of data storage for the business logic
• Caching, contains the Redis caching code, conforming to the business layer data storage abstraction
• Repo, contains the DB persistence code, also conforming to the business layer data storage abstraction

This layered system has the following interface layers

• Service interfaces
  The router layer demands implementations of the service layer
• Data storage interfaces
  The service layer, and caching layer, demand implementations of the data storage layer

By splitting it in these layers, it becomes possible to:

• Test the router JSON validity and request parsing in isolation
• Test the service behavior with easy to inject data types and analyze its behavior on the data storage in isolation
• Test the caching behavior in isolation
• Test the repo query logic in isolation
• Swap out the caching layer for direct repo access
• Swap out the repo implementation with another