I agree here. I understand the trade offs being made but I just don't see this being useful for too many people with zero fault tolerance and the single point

I agree here. I understand the trade offs being made but I just don't see this being useful for too many people with zero fault tolerance and the single point of failure meta data master. It probably gets some pretty good performance improvements from not worrying about that though.