Odoop

Potential features and improvements include in no particular order

• Multiplexer for map or reduce inputs so that the same data set can be processed in parallel (this may be difficult to manage when jobs need to be retried, etc.)
• Migrate HKey from internal library
• Migrate HBaseTemplate from internal library
• Add hbase shell support for query planner
• Add column constraints and boolean logic
• Ship ScalaObject along with hbase-filters
• Add unit tests to ensure garbage collection of deployed filters
• Add ability to register multiple filter classes with DeployedFilterManager and clean up at client shutdown
• Use distributed cache for jar shipping, so that HDFS does not get bogged down when the Scan starts

Documentation tasks include in no particular order

• Performance metrics
• Document code intent of query planner (i.e., write one line of SQL-like stuff instead of amassing multiple layers of FilterList and other filters)