SQL is one of the most widely used languages to access, analyze, and manipulate structured data. As Hadoop gains traction within enterprise data architectures across industries, the need for SQL for both structured and loosely-structured data on Hadoop is growing rapidly Apache Drill started off with the audacious goal of delivering consistent, millisecond ANSI SQL query capability across wide range of data formats. At a high level, this translates to two key requirements – Schema Flexibility and Performance. This session will delve into the architectural details in delivering these two requirements and will share with the audience the nuances and pitfalls we ran into while developing Apache Drill.