Efficiency of various compressors

The quantity of retained data is growing at a rate causing distress at technology centers. While only part of a solution, better compression methods are needed. The graph below shows compression of a 174.7 GiB tar file (my raw PhD research results) relative to the default gzip compression (gzip -6). The following are the shorthand names for the compression methods. The trailing number in the graph labels, where present, represents the preset level of compression. E.g. gz9 is equivalent to ‘gzip -9’. (The compression program zip (Info-zip) 3.0 was not included due to the default compression method being effectively identical to gzip.) Final size of the lzip -9 results: 1.4 GiB.

lz = lzip 1.10
xz = xz 4.999.9beta
gz = gzip 1.3.13
cmp = compress 4.2.4
bz = bzip2 1.0.5

The compression by lzip has an impressive advantage over gzip but at the expense of time. The compression times where not recorded, but it would be interesting to plot to show realtive cost of compression. In some cases, the compression may be seen as a one time event while disk space is an on-going expense. In some cases decompression time may be more of a concern.

A much more complete comparison of compression methods can be found at MaximumCompression.