In WinSrv 2012 you were forced (in a quite awkward way I must say) to predict what SSD cache sizes you needed and only using "Windows Storage Spaces" (WSP) and when done you were practically stuck with that configuration. And the internal cache management of WSP wasn't even real time since "hot" sectors had to be moved in and out by a background batch job running each night.
Actual requirements
What I'm looking for is a more general and true real time disk L2 caching using for example SSD that is not dependent of WSP thus could be used on top a any kind of drive or disk subsystems. Also, I want to be able to alter cache size and allocate space to different disks "live" i.e. without the need to reboot or deallocate the whole SSD-drive. Nice to have but optional is the ability to change between different cache algorithms like LRU/MRC/ARC etc.
Question:
Has this changed in anyway in WinSrv 2016 and can any of the above needs be implemented? If that’s the case, are there any architecture and design principles lying around somewhere that might describe how?
Thanks in advance "anyone"!