Weed-FS is more of a key-file mapping, without full POSIX support, and Weed-FS has less components. One file is identifiable via a 33 character string as ID.
All the file meta information is in memory, only one disk read is needed to get the file content, without too many disk operation to read file meta data.
You can call Weed-FS as NoFS, similar to NoSQL in the database world.
Frankly, I don't use other distributed file systems. All seems more complicated than necessary. Please correct me if anything here is wrong.
Compared to MogileFS: WeedFS has 2 components, directory server and storage nodes. MogileFS has 3 components, tracers, database, and storage nodes. One more layer means slower access, more complexity, more fail possibility.
Compared to GlusterFS: WeedFS is not POSIX compliant, and has simple implementation. GlusterFS is POSIX compliant, much more complex.
3 comments
[ 3.1 ms ] story [ 16.4 ms ] thread[1]: http://www.gluster.org/ [2]: http://danga.com/mogilefs/
All the file meta information is in memory, only one disk read is needed to get the file content, without too many disk operation to read file meta data.
You can call Weed-FS as NoFS, similar to NoSQL in the database world.
Compared to MogileFS: WeedFS has 2 components, directory server and storage nodes. MogileFS has 3 components, tracers, database, and storage nodes. One more layer means slower access, more complexity, more fail possibility.
Compared to GlusterFS: WeedFS is not POSIX compliant, and has simple implementation. GlusterFS is POSIX compliant, much more complex.