Why is thin-provisioning only supported with DDP?

[#1]

Greeting everyone,

I am currently very new to the great E-series and dynamic disk pool. To the best of my knowledge, the design of dynamic disk pool is adopted mandatorily or optionally in the solutions of other brands, say HPE 3PAR and Dell PowerVault MD-Series. However an interesting common thing is that thin-provisioning is only supported in DDP mode. That is, if raid group are arranged in traditional RAID 5/6, then volume or LUN within cannot be thinly provisioned. It is not unique to NetApp, but to Dell and HPE. Why is traditional RAID and thin provisioning so exclusive?

Any reply would be much appreciated. Thanks in advance.

[#2]

Hi,

In thin provisioning the system allows to provision storage beyond the physical capacity of the available storage. In case of total capacity of 10TB(10x 1TB drives) you can even create 20*1TB thin provisioned volumes. It is possible with logical volume manager.

Volume groups (VGs) can be resized online by absorbing new physical volumes (PVs) or ejecting existing ones. So if the volume group of 10TB is created, the maximum capacity it can provision is 10 TB. The mapping is 1:1 to the physical address space. 

So if we try to thin provision a volume group it will fail.

Refer http://www.tecmint.com/setup-thin-provisioning-volumes-in-lvm/ 

[#3]

Thanks for your reply. I know what thin provisioning is. In FAS-Series, thin provisioning for FlexVol can always be enabled despite the raid type of the underlying aggregate. My question is why in the e-series only under DDP can thin provisioning be supported. Does this feature have anything to do with the physical things? Thank you.

[#4]

E series have volume groups and DDP. In case of volume groups the physical space is managed such that every logical address needs a mapping to phsical block address. Since thin provisioning does not assign a physical address of the allocated space right away, it does not work in volume groups.

[#5]

I think the answer is Dynamic Disk Pools add a logical layer this can be used to keep track and manipulate unlike a tradition raid configuation which has a traditial disk raid geomtry using logical block addreses on the disks hard coded. 

Having thin volumes on traditional raid is probably possible if you use a file system for example VMFS can create thin VMDK's ontop of a tradition Hardware RAID but in this case the files system is keeping track of the logical blocks in the volume and expanding and adding them as neccesary.