> You'd solve more > of that by having the middle layer speak "raw device" underneath and be > able to sit on top of things like iSCSI (yes, really).
Back in ye olden days we called these middle layers "kernel" and "filesystem" and had that maintained by specialists.
Yeah. Unfortunately that turned out in a number of cases to be things like specialists that considered fsync unimportant, or dropping data from the buffer cache without errors.
But what I'm mainly saying is that if we want to run postgres on top of a block device protocol, we should go all the way and do it, not somewhere half way that will be unable to help most people. I'm not saying that we *should*, there is a very big if in that.