Hi,

we are starting to buy some large RAID arrays to store 50-100 TB of
video data. The data will later be processed by an application that
reads 2-3 video streams with ~16MB/s.

The storage will be attached to an Debian Etch Server (Kernel
2.6.18), the application/clients will access the data with samba. 

The RAID array will be split in smaller volumes and I want to create
xfs filesystems on these volumes with a maximum size of 4TB.

With this we would end with 20-30 filesystems, which I think would not
be very feasible. Thus I would like to have one virtual filesystem (or
data area) which is exported as one share by samba.

Nobody would access the org. filesystems directly, only the virtual
unionfs.

I did a very quick test with unionfs and a couple of directories that
I mounted to a directory. This worked exactly as I wanted and seems to
be the solution I'm looking for.

Is there anything that could be a problem in this scenario, any
limitations?

I found this old posting:
unified storage
http://thread.gmane.org/gmane.comp.file-systems.unionfs.general/1070

Does that mean that I can't write 12 TB of data to a unionfs with 3 fs x
4 TB?

lvm is no option, because I need to create the filesystem on top of
lvm. And I really don't want to use fs >4TB.

Ralf
_______________________________________________
unionfs mailing list: http://unionfs.filesystems.org/
[email protected]
http://www.fsl.cs.sunysb.edu/mailman/listinfo/unionfs

Reply via email to