This document introduces Howl, a table management service for Hadoop that aims to provide a shared schema, data typing system, and table abstraction across tools like Pig, Hive, and MapReduce. It discusses Howl's motivations of enabling collaboration, interoperability, and evolvability. The architecture takes elements from Hive and allows different tools to share data. Howl presents data stored in HDFS as tables that can be partitioned and with columns of specific data types. It provides loaders and storages for Pig and input/output formats for MapReduce. The roadmap includes adding write support for Hive, notifications, and other features to improve interoperability between tools processing data on Hadoop.