Case Study: Skybox Imaging achieves scalable, 24×7 global satellite imagery processing with Cloudera

Sneak Peek into Skybox Imaging’s Cloudera-powered Satellite System

Skybox Imaging is a startup building a low-cost constellation of high-resolution microsatellites to capture frequent imagery and video of the Earth. Their challenge was processing and fusing large volumes of raw, multi-image satellite data into usable products and making all that data quickly queryable and scalable as the number of satellites grows.

Skybox selected Cloudera CDH and built a Hadoop-based pipeline—wrapping native C/C++ image algorithms in a proprietary BusBoy framework so they run as MapReduce jobs—and use Puppet, Oozie, Hive and HBase to manage, orchestrate and publish results. This architecture keeps data on spinning disk for fast, exploratory analysis, supports on-prem and EC2 scaling, and enables data scientists to ask arbitrary questions and publish timely, scalable imagery products that drive a 24×7 sensor network.

Skybox Imaging

Oliver Guinan

VP Ground Software