I'm looking into HCI solutions, most of which appear to be built around RAIN (Redundant Array of Independent Nodes), caching or tiering of "hot" and "cold" data in the form of fast, expensive cache disks (SSD/NVMe) and cheaper, slower capacity disks (SATA/SAS).
With Microsoft's Storage Spaces Direct, an example they give in their Docs for a hybrid server node is 4 x 4 TB SATA drives fronted by 2 x 800 GB SSDs functioning as the cache (they suggest cache sized as 10% of capacity storage as a good starting point).
Would it be better for performance to have 8 x 2 TB drives? What about the other way, 2 x 8 TB SATA drives? This had a massive impact in the days of RAID, but unsure if this is important in software-defined storage in general.
More spindles will result in faster reads and writes as the load can be spread across more drives. Typically the recommendations you see are based on systems with limited drive bays. My home NAS can only hold 4 drives unless I find some 5.25-3.5 adapters to fit in a couple more. From your example I'd have no choice but to do the 4x4TB with 2x800GB SSDs for my optimal price/performance given my space limitations.