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