On 06/27/2012 05:04 AM, Boris Epstein wrote:
> Hello listmates,
>
> Has anybody attempted to run an XFS in a file mounted via -o loop? The file
> size is about 20 TB and it resides in a larger XFS filesystem residing on a
> disk. That filesystem's size is about 25 TB. What sort of degradation
> should one expect under that sort of scenario?
>
> The reason I am asking is that I am trying to run a MooseFS chunkserver in
> that filesystem-in-a-file and looks like I am getting some pretty bad
> delays.
Why would you do something like this? MooseFS depends on parallelism which
means you want many smaller chunkservers instead of a few big ones. Also
MooseFS requires some initial roundtrips when accessing files which makes
it unsuitable for storing/accessing lots of small files. Basically you
trade in latency for throughput (thanks to parallelism) which makes MooseFS
a candidate for storing and transfering large files.
What are you trying to accomplish and what exactly do you mean by "pretty
bad delays"?
Regards,
Dennis
_______________________________________________
CentOS mailing list
CentOS@xxxxxxxxxx
http://lists.centos.org/mailman/listinfo/centos
