DocStore: Document Database for MySQL at Facebook

14 April 2:20PM - 3:10PM @ Ballroom F

Experience level: 
50 minutes conference

Rate This Session

At Facebook, we have built infrastructure and expertise for running MySQL at scale. However, there are some costs to using MySQL as a storage solution when viewed from a developer perspective. MySQL requires rigidly defining the schema of your data, which is a cost to developers, especially early in a product cycle when one is unlikely to know what the schema should be. DocStore aim to give developers in need of storage a robust, flexible, easy-to-use solution that allows them to focus on writing their application rather than managing their storage. In this session, we will talk about our current work on adding JSON document support in MySQL/Innodb. We will cover the following topics. * A new native column type, document, for JSON data. * A new efficient binary representation of JSON, called FBSON, for document column storage. * Built-in JSON functions. * Document path expression (e.g. Doc.activity.timestamp) in queries. * How secondary indexes are supported on document paths. * Our use cases


Software Engineer, Facebook
Peng Tian is currently a software engineer on MySQL Engineering team at Facebook. Prior to Facebook, Peng worked on Web search at Microsoft and Yahoo!.
Software Engineer, Facebook
Tian is a software engineer at Facebook, working on MySQL database engine development. Before joining Facebook, Tian worked on Amazon Redshift, AWS’s data warehouse service.