Is your feature request related to a problem? Please describe.

As an operator or potential one, I don't know where to find information in one place about considerations such as:

• Component architecture (esp. operational perspective e.g. core needs a relational database which is used for registry only, serving instances need to talk to core instances, core and serving APIs are mostly stateless and horizontally scalable, etc.)
  • Feast's options for storage and compute engines—especially as support for more is added—and what Feast configuration parameters apply to each of them
• How configuration is managed in general, e.g. Spring profiles (ref Helm Chart Upgrades #458)
• How to control logging (related: Standardize logging libraries for Java components #277)
• What metrics are exposed and how
• How Feast uses Kafka topics currently
• Monitoring what Feast has last seen as the state of its managed jobs (ref Add job management APIs #302)
• Upgrading Feast
  • Considerations for re-deploying ingestion with active streaming jobs

Probably more, but that's a start.

Describe the solution you'd like

A section created that we can start from with an outline. By "section" I mean to distinguish tracks of documentation for what an end user—data engineer, data scientist, or software engineer in a client of the serving API—wants to know about using Feast's features in a deployment that's managed for them (or at least they're wearing that hat at the moment and not their DevOps one), versus operators who take care of said management.

Taking Kubernetes as a (more complex) example:

• Cluster Administration as a Concepts guide
• Administer a Cluster tree of task-oriented docs

It's doubtful we'd get get all of the above exhaustively documented in one swoop. Starting with a pebble should lower friction for contributors to extend it, I know I'll be glad to help fill things in as we prep our production-readiness.