> i can certainly see how this would be annoying from a > scalability standpoint, but how often are you changing user > storage limits? it would, however, be most convenient to just > have one huge-ass partition :). >
Annoying from a scalability standpoint? gimme a break. one huge filesystem is annoying from a scalablility standpoint. I run lots of big raids here, and I do *not* make one big partition on them, even the ones not on OpenBSD - It's the same principles behind the reasons I don't like to run services by scaling to one bigger and bigger machine. Inevitably you hit one wall or another, and scaling up when you hit that wall is horrificly expensive, recovering from a disaster on something so monolithicly big is horrificly time consuming and/or expensive. I like to scale services by adding machines and storage by adding partitions. Doing otherwise is a sign of inexperience. something your raid/san vendor will love to hear when they talk to you as they reach into their salescritter bag for their handy-dandy bottle of extra-sandy lube. Now I didn't say I didn't tie multiple partitions into the APPEARANCE of a big filesystem with NFS/AFS and whatever (which I do) I just don't make monster chiller horror native partitions which then need to be backed up and recovered in non-geologic time. -Bob