Case Study: Rubicon Project achieves real-time, sub-500ms interactive analytics at trillion-event scale with Imply

Modernizing Rubicon Project’s Analytics Stack for Programmatic Advertising

Rubicon Project, a leading digital advertising exchange, needed to modernize its Performance Analytics reporting platform as traffic exploded to trillions of ad and bid requests and MySQL no longer scaled. The company required strict accuracy, sub-second query performance for interactive exploration, and timely ingestion to support real-time decisioning. To meet these goals Rubicon Project selected Apache Druid as its analytics store and partnered with Imply, leveraging Imply’s suite of tools to replace the legacy stack and handle programmatic advertising scale.

Imply helped implement a Druid-based solution with roll-up/pre-aggregation, a mix of real-time and batch ingest, and tooling for ad-hoc slice-and-dice queries and monitoring. The production cluster now ingests over 2 TB/hour (more than a trillion events per day), serves thousands of users globally with average response times under 500 ms, and reduces stored raw data by 10–100x—delivering interactive analytics, faster customer decisioning, and a more cost-efficient, scalable platform.

Rubicon Project

Ken Lin

Senior Director of Product Management