This repository demonstrates a data engineering pipeline using Spark Structured Streaming. It retrieves random names from an API, sends the data to Kafka topics via Airflow, and processes it with Spark Structured Streaming before storing it in Cassandra.

System Architecture

Components:

Data Source: Uses the randomuser.me API for generating user data.
Apache Airflow: Orchestrates the pipeline and schedules data ingestion.
Apache Kafka & Zookeeper: Stream data from PostgreSQL to Spark.
Apache Spark: Processes data in real time.
Cassandra: Stores the processed data.
Scripts:

kafka_stream.py: Airflow DAG script that pushes API data to Kafka during 2 minutes every 1 seconds.
spark_stream.py: Consumes and processes data from Kafka using Spark Structured Streaming.

Technologies:
Apache Airflow, Python, Apache Kafka, Apache Zookeeper, Apache Spark, Cassandra, PostgreSQL, Docker

Getting Started

WebUI links

Airflow : http://localhost:8080/
Kafka UI : http://localhost:8085/ \

Clone the repository:

$ git clone https://github.com/akarce/e2e-structured-streaming.git

Navigate to the project directory:

$ cd e2e-structured-streaming

Create an .env file in project folder and set an AIRFLOW_UID

$ echo -e "AIRFLOW_UID=$(id -u)" > .env

$ echo AIRFLOW_UID=50000 >> .env

Run Docker Compose to perform database migrations and create the first user account

$ docker-compose up airflow-init

Run Docker Compose again to spin up the services:

$ docker compose up -d

Copy the dependencies.zip and spark_stream.py files into spark-master container

$ docker cp dependencies.zip spark-master:/dependencies.zip

$ docker cp spark_stream.py spark-master:/spark_stream.py

Run the docker exec command to access cqlsh shell in cassandra container

$ docker exec -it cassandra cqlsh -u cassandra -p cassandra localhost 9042

Run describe command to see there are no keyspaces named in cassandra instance

cqlsh> DESCRIBE KEYSPACES;

Unpause the dag user_automation using Airflow UI

Go to Airflow UI using : http://localhost:8080/

Login using Username: admin Password: admin

You can track the topic creation and message queue using the open source tool named UI for Apache Kafka that is running as a container, WebUI link: http://localhost:8085/

Message schema looks like this

In a new terminal run the docker exec command to run spark job to read the streaming from kafka topic:

$ docker exec -it spark-master spark-submit --packages com.datastax.spark:spark-cassandra-connector_2.12:3.5.1,org.apache.spark:spark-sql-kafka-0-10_2.12:3.5.1 --py-files /dependencies.zip /spark_stream.py

Now go back to the cqlsh shell terminal back and run the command to see data is inserted to cassandra table called created_users

cqlsh> SELECT * FROM spark_streaming.created_users;

and run count query several times to approve data is being inserted while running user_automation dag

cqlsh> SELECT count(*) FROM spark_streaming.created_users;