Since I've migrated my NS480 (CX4-480) to Flare 30 version, I'm really studying to approach Clarioon pool for thick lun. Considering that I've used for a long time Metalun (especially for vmfs) and I'm really happy of using them, Clariion pools (thick Luns) seem to be very similar and very reliable (and seems that EMC is deprecating Metalun statement and pushing up Provisioning) but not as dynamic as Metalun ( see https://community.emc.com/thread/111534?start=0&tstart=0 )... So why choose It instead of Metalun if I need performance and dynamism?
in the same array, you can have pools to simplify admistration and raid group to keep performance.
So you can have thick and thin lun in the same array
Also you can migrate from thick to thin LUN and reverse.
So, if you have enough disk you can try in your environnment...
My experiency :
generally one pool contains some Raid Group Raid 5 4+1 (hidden for customer)
The problem is : in a pool you don't know where is LUN and you can have some hot spot inside your pool so poor performance...
In a raid group, you know where is all component LUN and you know if you have hot spot or not.
For me, the best is to have the two architecture :
Pool and Raid Group to have LUN easily manageable on Pool and Performance LUN on Raid Group.
if you want to improve performance, i recommand FAST Cache.
if you want to have lowest TCO, create Pool with different type of disk (SSD, FC, SATA) and use FAST Licenced (not for performance but for tco).
Remember : inside an array, you work with cache (very fast) and only after with disk (slower)...
I would use both technologies - both have there place I would use MetaLuns when I want to guarantee performance - pools have there place but require more planning and design.
You might want to review the EMC CLARiiON MetaLUNs: A Detailed Review whitepaper. The 'MetaLUN Alternatives' section describes the tradeoffs between MetaLUNs and pools. EMC CLARiiON MetaLUNsis available on Powerlink.