Origins of S3 and data lakes

How AWS S3 is built

Topic 2 of 13

Published Jan 21, 2026 · Clip 3:58–9:45 (5:47)

The guest recounts S3's origins in 2005, launched in 2006 as the first AWS service to store unstructured data economically. The original design was eventually consistent, optimized for durability and availability. She traces the evolution of data lakes: frontier customers like Netflix and Pinterest combining Hadoop with S3 (~2013–2015), enterprise adoption of parquet data (~2015–2020), the rise of Iceberg, and the 2024 launch of S3 Tables and S3 Vectors. She explains why Iceberg's decentralized analytics architecture is so popular.