/rubix

Cache File System optimized for columnar formats and object stores

Primary LanguageJavaApache License 2.0Apache-2.0

RubiX

[Build Status] codecov

RubiX is a light-weight data caching framework that can be used by Big-Data engines. RubiX uses local disks to provide the best I/O bandwidth to the Big Data Engines. RubiX is useful in shared storage architectures where the data execution engine is separate from storage. For example, on public clouds like AWS or Microsoft Azure, data is stored in cloud store and the engine accesses the data over a network. Similarly in data centers Presto runs on a separate cluster from HDFS and accesses data over the network.

RubiX can be extended to support any engine that accesses data using Hadoop FileSystem interface via plugins. There are plugins to access data on AWS S3, Microsoft Azure Blob Store and HDFS. RubiX can be extended to be used with any other storage systems including other cloud stores

Check the User and Developer manual for more more information on getting started.

Supported Engines and Cloud Stores

  • Presto: Amazon S3
  • Spark: Amazon S3
  • Any engine using hadoop-2 e.g. Hive can utilize RubiX. Amazon S3 is supported

Resources

Documentation
Getting Started Guide
User Group (Google)

Talks

Talk on Rubix at Strata 2017

Blog Posts

Developers

Slack Channel

The channel is restricted to a few domains. Send an email on the user group or contact us through Github issues. We will add you to the slack channel.