Unclear Docs: Server Section

[dan-jan]

Current Issues

• Server section is lacks conceptual clarity, and setup process is convoluted

Todos

• Pena team to edit
• Resource person: @hiento09 @hiro-v
• Approver: @dan-jan

Recommended Structure

• Server
  • Docker
  • Kubernetes
  • AWS
  • GCP
  • etc

Server (/docs/server)

• Page to be added later (@dan-jan to draft)

Server -> Docker (/docs/server/docker)

## "Compatibility"
	- Hardware
		- CPU
		- RAM
		- GPU (Optional) 
	- OS
## Prerequisites
	- Step 1: Install Docker Engine and Compose
	- Step 2 (Optional): Install GPU Drivers
		- Nvidia
		- Intel
		- AMD
## Installing Jan
:::note
- For this section: we go with the defaults (e.g. GPU, and with local Data Folder). 
- We should note there are alternatives, and link the other other sections in "Advanced use"
:::
- Step 1: Download and AmendJan's Docker Compose file
	- Go through Docker Compose file, and explain it
	- Refer to the "Dockerfile, Explained" section in the Resources section (at end)
	- Updating Dockerfile with CUDA version
- Step 2: Run Jan Docker Compose with your variables
	- NOTE: runs with local data folder. If they want autoscaling, recommend them use Remote Data Folder (or Kubernetes)
	- GPU mode (recommended) or CPU mode (can 
		- `docker compose`
		- Explain what it does behind the scenes
		- CPU mode 
	-Step 3" Access Jan Server

## Advanced Use

### Remote Data Folder
- Explain how this enables autoscaling (e.g. multiple Docker containers -> pointing to single Data Folder)
- Remote S3 bucket
- `gpu-s3fs` and `cpu-s3fs` (can consider using tabs for this)
- Recommended for Autoscaling, etc
- Recommend running with k8s if this is your usecase

### CPU use
- Talk about how this isn't super recommended as it's slow
- Provide how it can be used here

## Resources

### Understanding Jan's Dockerfile

- What are the different variables?
- Diagram of different Jan Containers

Server -> Docker (/docs/server/kubernetes)

## "Compatibility"
	- Hardware
		- CPU
		- RAM
		- GPU (Optional) 
	- OS
## Prerequisites
	- Step 1: Install Docker Engine and Compose
	- Step 2 (Optional): Install GPU Drivers
		- Nvidia
		- Intel
		- AMD
## Installing Jan
:::note
- For this section: we go with the defaults (e.g. GPU, and with local Data Folder). 
- We should note there are alternatives, and link the other other sections in "Advanced use"
:::
- Step 1: Download and Amend Jan's Docker Compose file
	- Go through Docker Compose file, and explain it
	- Refer to the "Dockerfile, Explained" section in the Resources section (at end)
	- Updating Dockerfile with CUDA version
- Step 2: Run Jan Docker Compose with your variables
        - Note: Default to using the remote Data Folder, recommend Docker if they want to use a local Data Folder
	- Docker compose command with `gpu-s3fs` or `cpu-s3fs` (can consider using tab for this)
	- GPU mode (recommended) or CPU mode (can consider using tab for this)
		- `docker compose`
		- Explain what it does behind the scenes
		- CPU mode 
	-Step 3" Access Jan Server

## Advanced Use

### CPU use
- Talk about how this isn't super recommended as it's slow
- Provide how it can be used here

## Resources

### Understanding Jan's Helm chart

- What are the different variables?
- Diagram of different Jan Containers