[TAJO-49] Block-based compression of RowFile

• Type: New Feature
• Status: Resolved
• Priority: Major
• Resolution: Won't Fix
• Component/s: Storage
• Labels: gsoc gsoc2013 mentor
• Assignee: Unassigned
• Reporter: Jihoon Son
• Created: 29/Apr/13 07:05
• Updated: 15/Oct/15 04:38
• Resolved: 15/Oct/15 04:38

Description

Background

RowFile is the row-oriented binary file format of Tajo. Currently, RowFile does not support compression.

Proposal

The goal of this proposal is to implement block-based compression of RowFile. Requirements are follows.

• The compression type should be configurable via TableMeta.
• The block size also should be configurable via Configuration.
• LZO, BZip2, GZip should be supported. (I recommend utilizing Hadoop's CompressionInputStream/CompressionOutputStream)

Issue Links

• is related to: TAJO-134 Support for compression/decompression of CSVFile