Case Study: Iflix achieves real-time streaming analytics and faster ETL with Qubole

Iflix Distributes Video Streaming Online With Real-Time Analytics And Recommendations

Iflix, a Malaysia-based OTT service with operations in 28 countries and over 10 million paid subscribers, faced growing data- and ML-driven demands as user activity and content offerings expanded. Their original pipeline routed event data to Kinesis and S3 but relied heavily on Redshift for ETL and reporting, leading to long batch windows (6–10 hours) and limited ability to deliver near-real-time analytics and recommendations at scale.

To address this, iflix decoupled storage and compute, keeping raw data in S3 and adopting Qubole’s cloud big-data platform with Spark (orchestrated via Airflow), Presto, and a separate VPC for role-based access. This new architecture cut ETL time by about 40% for a sample stream, enabled much faster analytics and experimentation, and used autoscaling with AWS Spot instances to improve cost efficiency and resource flexibility for engineers, data scientists, and analysts.

Iflix

Bruno Gagliardo

Global Director of Data Analytics