GCPAZURE Deep Dive
Pipeline Orchestration & ETL

Cloud Composer (Apache Airflow) Azure Data Factory (ADF)

From Cloud Composer (Apache Airflow Python DAGs) to Azure Data Factory (Visual Serverless Pipelines).

The 30-Second Mental Model Shift

Cloud Composer requires maintaining a GKE cluster running Airflow workers in Python ($250+/month baseline even when idle). Azure Data Factory is 100% serverless ($0 baseline) with a visual drag-and-drop canvas and the Self-Hosted Integration Runtime that acts as an outbound-only secure gateway to your private corporate databases.

1. Architectural Mechanism Comparison

GCP (What You Know)
Source

Cloud Composer (Apache Airflow)

Managed Apache Airflow where 100% of pipeline DAGs, operators, and dynamic task graphs are authored in Python code.

Key Architecture Strengths:
  • Complete programmatic flexibility using Python code and custom operators.
  • Rich Airflow open-source ecosystem, plugins, and custom hooks.
  • Version-controlled Git workflows for all DAG definitions.
AZURE (How It Works)
Mastery Target

Azure Data Factory (ADF)

Serverless visual ETL and orchestration service backed by JSON schemas. Connects to 100+ native sources with built-in Self-Hosted IR for hybrid on-prem networks.

Why Azure Built It This Way:
  • 100% serverless ($0 idle cost vs. Airflow GKE cluster baseline).
  • Self-Hosted Integration Runtime (SHIR) securely connects on-prem with outbound-only port 443.
  • Mapping Data Flows executes visual transformations on serverless Apache Spark.

2. Interactive Terminology & Concept Bridge

Interactive Concept Bridge: Terminology & Architectural Mapping

Click any concept below to see how your GCP knowledge directly maps into AZURE.

Mapping Deep Dive
Similar Mechanism
⚡ Direct cognitive shortcut
GCP (What You Know)

Airflow DAG (Python File)

Directed Acyclic Graph defining tasks and execution dependencies.

AZURE (How It Works)

ADF Pipeline (JSON Schema)

Visual workflow grouping activities with On-Success/On-Failure links.

The Architectural Mental Shortcut:

Both represent end-to-end data workflows with dependency management.

3. Visual Architecture Pipeline (Azure Data Factory (ADF))

Azure Data Factory (ADF) 3-Stage Architecture: Ingest ➔ Integration Runtime & Spark ➔ Sinks

Click any section below or run the simulation to see how ADF orchestrates hybrid data pipelines.

1. Linked Services
2. Integration Runtime
3. Target Sinks
Connectors
100+ Native
On-Prem Security
Outbound 443
Transform Engine
Serverless Spark
Billing Metric
DIU / Sec ($0 Idle)
The Engine
Stage Details

2. Integration Runtimes (IR) & Mapping Data Flows

The Integration Runtime executes pipeline activities. Azure IR manages serverless cloud movement; Self-Hosted IR securely connects private on-premise networks with outbound-only HTTPS (port 443); Mapping Data Flows compiles visual graphs to Apache Spark.

Real-World Analogy

Like a secure armored courier that can enter high-security bank vaults (on-prem) without leaving the back doors unlocked.

Key Mechanics
  • Self-Hosted IR (SHIR): Outbound-only port 443 (zero inbound firewall holes required).
  • Azure IR: Managed serverless cloud compute auto-allocating Data Integration Units (DIUs).
  • Mapping Data Flows: Drag-and-drop transformations compiled into Spark Scala jobs.
Hybrid Security
Outbound 443
Data Flow Engine
Serverless Spark

4. Side-by-Side Code, CLI & Terraform Translator

Side-by-Side Code & Syntax Translator

GCP Syntax
# Cloud Composer (Airflow) CLI Trigger
gcloud composer environments run my-airflow-env \
  --location us-central1 \
  dags trigger -- DailySalesETL
AZURE Equivalent
# Azure Data Factory CLI Trigger
az datafactory pipeline create-run \
  --factory-name adf-corp-prod \
  --name DailySalesETL \
  --resource-group rg-analytics
Code Translation Notes:Both CLI commands trigger an immediate execution run of the workflow.

5. Paradigm Shift Gotchas: Traps to Avoid in AZURE

Gotcha #1
tip

The Self-Hosted IR Outbound Rule

The Trap:

Network teams often assume Azure needs inbound access to on-prem databases and deny requests.

How to Avoid It:

Clarify to security teams that SHIR uses outbound-only port 443 to poll Azure for jobs.

6. Test Your Mental Model

Quick Knowledge Check: Test Your AZURE Mental Model

Solidify your cross-cloud understanding with instant feedback.

1How does Azure Data Factory execute visual Mapping Data Flows without requiring the user to write code?