Case Study: Zillow achieves reliable, cost-efficient, scalable Presto and Spark analytics with Qubole

Zillow Powering The Complete Online Home Lifecycle With Qubole

Zillow, the online real estate marketplace, needed to consolidate growing volumes of brand and application data into an S3-based data lake while supporting thousands of daily users and hundreds of analysts running terabyte-scale Presto and Spark queries. The Data Infrastructure team faced inefficient autoscaling and Spot instance buying, frequent Presto instability, and high costs and operational overhead for large daily clickstream jobs, all while operating with a small team.

Zillow adopted Qubole’s cloud-native data platform—leveraging workload-aware autoscaling, aggressive downscaling, automated Spot buying and expert support—to stabilize Presto and optimize Spark processing. The result was dramatic cost savings, 25–50% greater infrastructure team productivity, reliable clusters that handle large and variable workloads (including ~3 TB/day clickstream jobs), fewer missed SLAs, and improved trust in query results.

Zillow

John Cruchon-Dupeyrat

Principal Product Manager for Big Data