Recommended iSCSI LUN size for creating a ZFS pool on a Solaris 10 client ?

I'm not a Clariion expert thus can't talk about lun size on CX.

But generally speaking you use external storage to offload host from storage tasks (just like protecting data). This usually saves you CPU power you can use elsewhere. Eventually to run your database engine

if you are presenting iSCSI LUNs from Celerra, they have already been striped twice, once when your raid group was created, and second time when you created file system for iSCSI LUN ( it was created from one of the system defined pools like clarata_archive ? )

Correct, but does this direct one towards LUNs as large as possible
or is there a certain optimal value ?

Furthermore, when opting for the self-healing properties of ZFS and
going the RAIDZ route, I'll suffer a capacity penalty of one LUN's
worth.

I'll rephrase my original question : is 11 LUNs x 400 GB in RAIDZ a
bad idea for creating a 4 TB Zpool ?

Is 6 x 800 GB a better idea ?
Is 21 x 200 GB better ? (gut feeling says : not)

FYI, the Solaris 10 server(s) is/are not really short on CPU cycles.


Eric.

Eric even if your servers are not really short on CPU cycles, they need to waste a few of their cycles to protect your data. But the Celerra is already protecting them.

I guess ZFS self healing properties are good with a JBOD. Unfortunatly a Celerra isn't a JBOD