Iceberg with Flink at DoorDash

#icebergSummit 2025 breakout session delivered by Tristan Culp and Gaurav Sharma from #DoorDash Session Description: In the modern era of data-driven decision-making, there is a growing need for highly scalable and robust data platforms, capable of ingesting data in near real time. At DoorDash, we integrated Apache Iceberg with Flink to scale our data ingestion solution, handling millions of events per second into our data lake to support near real time data consumption. We’ll share why we chose Iceberg, and how migrating our table format lowered our storage and operational costs while increasing our data reliability. In this talk we’ll also discuss some of the challenges we faced in its adoption, such as performance issues and limitations with the Flink Iceberg sink. We’ll cover the changes we made to address these hurdles, how we plan to further integrate Iceberg into our architecture in the future, and practical insights on operating streaming Iceberg at scale. Join us for a deep dive into how we are productionizing Iceberg at scale, and the strategies that have helped us succeed at DoorDash.