built-in rotating and compressed files: rotfs and zfs
Many applications make heavy usage of compressed files and some need rotating files.
Though compression methods are almost stabilized as standards, one still does need special tools like zcat to read from them. But many applications are not aware of zcat, zcat cannot read byte ranges within binary files ...
Hence a proposal: how about implementing a file compression and a file rotation within the kernel ?
- This would make system calls like read() and write() transparent ... so it would brought transparency to all applications
- e.g. syslog messages would be compressed on-the-fly, backup utilities would not have to delay job queues due to compression overcost.
- Humans would not have to bother about disk usage and machine power when uncompressing files.
So many advantages !
Now, one open question: is it better to implement this at file level (one extra attribute) or at file system level (designing zfs and rotfs) ?
Regards
Phil
Look at FiST, which is available from http://www.filesystems.org. The FiST package includes gzipfs, a stackable file system that compresses and decompresses data on the fly.
Charles
Quoted 29 lines · On Thu, 2004-01-22 at 10:34, [email protected] wrote:
Many applications make heavy usage of compressed files and some need rotating files.
Though compression methods are almost stabilized as standards, one still does need special tools like zcat to read from them. But many applications are not aware of zcat, zcat cannot read byte ranges within binary files ...
Hence a proposal: how about implementing a file compression and a file rotation within the kernel ?
- This would make system calls likeread()andwrite()transparent ... so it would brought transparency to all applications
- e.g. syslog messages would be compressed on-the-fly, backup utilities would not have to delay job queues due to compression overcost.
- Humans would not have to bother about disk usage and machine power when uncompressing files.So many advantages !
Now, one open question: is it better to implement this at file level (one extra attribute) or at file system level (designing zfs and rotfs) ?Regards
Phil
-
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html