1. PCIe adapters: Requires free PCIe slot, but most systems have limited number. Best performance and low cost, but it would leave a lot of modules on the table.
2. USB3 adapters: Allows expanding systems without PCI, but I can't find adapters that will safely a 22110 module. Less performant, but allows expanding storage almost indefinitely. Still, a relatively low cost option, even if I include the need to design and print suitable custom enclosures. Note: this does not seem to allow low-level access to the M.2 drive using `nvme`, so it may not be possible to upgrade the firmware for these drives.
3. COTS NAS: Safely utilizes numerous modules in one place with the least effort and disruption to existing systems. Sadly, I have not found an off-the-shelf option that supports 22110 M.2 modules, though this should be a high performance solution (if I'm willing to eat the predictably higher cost).
4. Custom NAS: I could do the initial design work (e.g. board layout and modeling), attract others with complementary skills, and then launch a crowdfunding campaign that aims to build a suitable product in order to serve this niche market.
As the last should suggest, there are no bad ideas here, just limited resources to pursue them.
What would you do, fellow HN reader?
NVME drives seem to be cheaper than I remember - you might be holding a depreciating asset.