From mboxrd@z Thu Jan 1 00:00:00 1970 From: PFC Subject: Re: Compression Plugin Date: Wed, 21 Sep 2005 10:33:30 +0200 Message-ID: References: <433017C1.4050408@gmail.com> <43307E49.8060404@slaphack.com> <433102D0.7010706@mch.one.pl> Mime-Version: 1.0 Content-Transfer-Encoding: 8bit Return-path: list-help: list-unsubscribe: list-post: Errors-To: flx@namesys.com In-Reply-To: <433102D0.7010706@mch.one.pl> List-Id: Content-Type: text/plain; format="flowed"; delsp="yes"; charset="us-ascii" To: reiserfs-list@namesys.com >>> An interesting idea: select the algo and a range of compression >>> levels per file, A simple check on wether it's an already compressed file (using file extension and magic number) should be quite easy to do and cheap. Now, intrigued by this lzo thingie, I ran a little benchmark on my emails ; I have a huge mail spool and my mail client is always slow to launch... Seems that on this laptop LZO can compress about 40 MB/s and decompress about 200 (!) MB/s. On the mail spool, compression ratio was about 1/2 ; gzip was better, although 5-6 times slower. So considering the disk throughput is only 15 MB/s, this could make it twice as fast with CPU to spare. Yowza ! >>> but select the actual compression level at flush time >>> based on some estimate of how loaded the system is.. :) >>> Probably not worth it even though the amount of compression and the >>> speed differ greatly from -1 to -9... I hope no one wastes their time >>> on it until the more important things are done.. but perhaps a nice >>> touch. >> Kinda makes it sound like in addition to the "repacker" that repacks >> blocks in order and squeezes them, (which should allow for filesystem >> size reduction), there'll be a compression repacker too that allows >> you to repack files to a higher or lower compression level (anyone >> remember the Compression Agent that came with DoubleSpace and the like >> in Windows 9x?). > > AFAIR e2compr patch to ext2 (which enables compression on a ex2 > filesystem) could use many packing methods, including gzip and bzip2 > (!), with a range from 1 to 9... > > I certainly could imagine cases where compression level is more > important than speed. > >