Details
-
Improvement
-
Status: Closed
-
Major
-
Resolution: Fixed
-
0.11
-
None
-
None
Description
The default sync interval is 16K which is very bad for bzip compression which can take bigger chunk of data for compression. Hadoop's Bzip2code uses io.file.buffer.size as the buffer size. Most tuned environments have it set to 128K which gives better compression.