Details
-
Improvement
-
Status: Closed
-
Major
-
Resolution: Fixed
-
0.5.0
-
None
-
None
Description
In hadoop streaming, a record is represented as a String for I/O and is encoded as UTF8 for map/reduce. A record has to be converted between String and UTF8 back and forth multiple times and this wastes CPU time.
Attachments
Attachments
Issue Links
- incorporates
-
HADOOP-413 streaming: replace class UTF8 with class Text
- Closed