Introduction
I was struggling recently with a Windows Storage Spaces problem with a parity virtual disk. The nature of my problem is not important to this discussion, but rather it is the fact that during my fruitless pursuit of a solution, I got a lot of things “explained” to me (indirectly) by self-appointed authorities decrying the use of software parity RAID (traditionally RAID-5, -6, -50, and -60) and even software mirroring. Other experts — while acknowledging certain limitations — defended software RAID, often referring to the ubiquity and reliability of software RAID solutions available on modern platforms and even going as far as saying that software is superior because it avoids the single point of failure that is the hardware RAID controller itself. (Spares are unavailable sometimes, I guess.)
All this free online haranguing got me thinking about what RAID-5 performance is really like given a specific system. That is, if I had a certain computer and hard drives, how would software methods perform compared to hardware for parity-redundant storage? And so, I got a certain computer and hard drives and tested it. Here is my story.