Details
-
New Feature
-
Status: Closed
-
Major
-
Resolution: Not A Problem
-
0.92.0
-
None
-
None
-
None
Description
Not sure if I'll have time to work on this, but the old sequencefile-based import/export are not as efficient as one based on HFiles would be. Idea is:
- hbase exporthf: TableInputFormat -> Identity -> HFileOutputFormat (should also take time ranges, etc, for incremental backup)
- hbase importhf: perhaps this is just an improvement to the existing completeBulkLoad that does parallel splitting when the input HFiles don't "fit" the target regions (right now this is single-threaded)
This would be a good project for a new contributor!