A large number of small luns or single large lun size for VxVM on NetApp

[bug]

Hi experts,

     As i known, if would like to perform better performance then we must create VxVM as stripe layout to spread all IO across disks or luns but in case of NetApp storage array if luns configured under same aggregate volume then IO will automatically distribute to all disks in aggregate volume, so i want to know what recommend configuration if we use VxVM on NetApp storage? or any best practice? use the single large lun size (distribute IO by Storage array) ?or number of small luns (distribute IO by VxVM first then do again by Storage) ?

pinyo