• UI (Web Server)
• Scheduler (like project manager). Work with Executor. Run DAGs
• Workers (processes)

Additionaly run SQLite.
Noone use SQLite in prod because it one-thread and not robust

In prod we can use PostgreSQL, MySQL, MSSQL

By default Airflow creates SQLite DB
airflow standalone - config and create DB for the first time

Directed acyclid graph
Take args to initialize

Simple code:

DAGs starts in start_date + interval
start_date must be in the past

Need to create connection
Use it for passwords and confidectial info that do not for commom user

Automize

• Tracking
• Projects (packing experimantations)
• Models (Packing models)
• Registry (repository)

S2 - Simple Storage Service
S3 for AWS initially
We can access for DB using URL and locate Models in it
Connection realize using API or Web-interface