On 8/12/06, Ken Snider <ksnider at flarn.com> wrote: > Consider my vote for this as well. :) The reality is, with insert speed the > way it is, inserting as little data as possible should be a goal, and while > bzip2's compression method is without question slower than gzip, I'm sure > we'll be blocked by insert velocity long before the compression rate becomes > an issue.
Please no bzip2. The problem with bzip2 is that it is every bit as slow to compress as it is to decompress. If better compression is required please look at getting someone to do a java port of LZMA. LZMA is in generally somewhat faster than bzip2 for compression (and gets better compression rations on most content, to much better compression on some content) and is MUCH faster than bzip2 for decompression.
