Hi sage, Thanks for explanation. We read ceph papers and known ebofs has good performance by writing the data to device directly. Does btrfs have same performance as ebofs?
Thanks, Guiwen >Hi, >> >We dropped support for ebofs a few months back since the plan moving >forward is to use btrfs and the maintenance burden wasn't deemed >worthwhile. Btrfs does (almost *) everything ebofs does, and much more, >and is very actively developed. >The main issue with building ebofs at this point is pobject_t. It used to >be a simple wrapper around object_t, which used to be fixed-size. >object_t is now essentially a variable length string, which won't work >with ebofs' btree code. Either that code needs to be rewritten >accordingly, or you can redefine object_t as a fixed size struct and use >ebofs outside of ceph. There are a variety of places now where ceph uses >variable-length object names that would need to be cleaned up to use ebofs >on a ceph osd... >sage On Tue, 18 Aug 2009, ºî¹ðÎÄ wrote: > Hi, > > We are interesting EBOFS. It seems there are 3 types of OSD in Ceph, > filestore, ebofs, bdb. Currently the default setting is filestore, that > means on regular file system such as btrfs , ext3 etc. We want to test the > system on ebofs. Is there any document on how to setup the system by using > ebofs? > > > > Thanks > > Guiwen > > > > > > ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Ceph-devel mailing list Ceph-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/ceph-devel
