[SQOOP-1532] Sqoop2: Support Sqoop on Spark Execution Engine

• Type: Improvement
• Status: In Progress
• Priority: Major
• Resolution: Unresolved
• Fix Version/s: 2.0.0
• Assignee: Veena Basavaraj
• Reporter: Veena Basavaraj
• Created: 18/Sep/14 21:48
• Updated: 15/Dec/15 15:46

Description

The current execution engine supported in sqoop is MR.

The goal if this ticket is to support sqoop jobs ( map only and map+reduce ) to run on spark environment.

It should at the minimum support running on the standalone spark cluster and then subsequently work with YARN/mesos.

High level goals
1. Hook up with the connector apis to provide the basic load/ extract to the spark RDD.
2. Implementation of the Sqoop RDD to support extraction from different data sources . The design proposal will discuss the alternatives on how this can be achieved.
3. Optimizing the loading/writing ( re-use/ refactor the consumer thread code to be agnostic of the hadoop output format)