On Mar 14, 2006 12:21 -0500, Ron Gage wrote:> 1. Is there any redundancy mechanism that works with Lustre such
> that if a storage object goes down for whatever reason, we don''t
loose
> any data - something akin to RAID-5. Yes, I understand about the write
> avoidance mechanism that takes into account if the storage object is
> alive or not, I am more concerned about not losing access to data.
There is currently no support for lustre-level data redundancy, though
work is underway to implement it. It is possible to have this at the
hardware level for some types of storage systems, and others have
recently reported some success with DRDB (this does block-device level
mirroring between hosts).
> 2. Does Lustre play well with SAMBA?
It is my understanding that it does, though we don''t test this
ourselves.
> 3. Is anyone working on a "Windows Client"?
There has been some work in this. If this is something you are interested
in then you should contact sales@clusterfs.com to voice your interest.
Cheers, Andreas
--
Andreas Dilger
Principal Software Engineer
Cluster File Systems, Inc.