What I know and what I prefer are the result of personal decisions I made after leaving my financial/IT work-life for retirement in 2000. I was motivated to provide some sort of household server, that would also allow installation of "server" software versions, like DBMSes. That in turn allowed me to teach these aspects of topics to others. But a simple peer-to-peer server, using either workstation Windows or server Windows, was inexpensive because I could use older parts as I upgraded to newer processors, chipsets, etc.
I've used RAID0 and RAID5. I've always had data backed up in some simple way, although the backup intervals were too far and few between. But since my server is not accessed by thousands of people simultaneously, there is a departure in the size of my investment, the number of disks, the lack in use of Xeon and Super-Micro server boards. And ultimately, the bottleneck as I see it resides with the speed of my local Ethernet LAN, or Gigabit. At that point, it would seem meaningless to open the storage bottleneck any wider with two dozen hard disks and $1,000 controllers.
And so I dispense with the idea of upper-tier hardware controllers with large DIMM buffers, PCI_E x8 bandwidth, as examples. I found a simple SuperMicro 8-port controller that only provide JBOD and ACHI, connecting SATA-III drives with the 8087 breakout cable and two plugs on the controller card for two cables. I think I spent about $110.
And I chose to put these drives into a software-managed drive-pool, with function provided for data-duplication/redundancy at the file and folder level. I cache the drives to a combination of RAM and SATA SSD drive-space, but at most this only brings full saturation to match the gigabit network speed.
I don't know if drivepool configurations can be implemented on NAS devices, but these NAS devices have their own operating system or there are open-source systems.