Why Doesn’t ADPv2 Support HDDs?

[kevin963210]

I’ve noticed that NetApp ONTAP’s ADPv2 (root-data-data partitioning) is only supported on SSDs (AFF systems), but not on HDD-based FAS systems.

Does anyone know the technical reasons behind this limitation?
Are there any official documents or references that explain why ADPv2 cannot be applied to HDDs?

[TMACMD]

Spinners don’t have the bandwidth. 
SSDs have lots of bandwidth and can handle the extra load. The imagine a spinner trying to jockey have two different nodes send the read/write heads all over at the same time. 
i would equate it to trying to copy data off a cdrom in two different windows. Back and forth and back and forth

[chamfer · Accepted solution]

@kevin963210 building on what @TMACMD has advised the reason that ADPv2 is not supported on HDD or mixed SSD/HDD configuration is due to the lack of IOPS (Input/Output Operations Per Second) of HDDs.  HDDs have an IOPS of approx. 150 per drive, whereas SSD's are > 10,000 IOPS per drive (NVMe drives can be 500k IOPS).

With ADPv2 on HDD you would terrible performance with the two data aggregates stripped across a drive when both aggregates were heavily active, unlike ADPv1 which has a single root and single data partition per drive. 

ADPv1 does not have data partition contention like ADPv2 but is not a problem with AFF due to the high IOPS of flash.

In AFF systems you generally run into controller limitations before you run into drive IOPS limitations.