This document summarizes Spark's development over the past 12 months and provides a look ahead. It discusses improvements to both the frontend, such as DataFrames and machine learning pipelines, and the backend through projects like Tungsten for performance optimizations. Going forward, it mentions new features like the Dataset API, streaming DataFrames, and potential hardware improvements from technologies like 3D XPoint memory. The overall goal is to provide a unified engine and APIs that can automatically optimize analytics workloads across languages and domains.