We bypassed HDFS and connected Apache Spark to Google Cloud Storage

blog.datalicious.com