begin quoting Neil Schneider as of Sun, Dec 16, 2007 at 06:07:57AM -0800: > > SJS wrote: > > > That's exactly what I mean. Requiring a driver renders the "otherwise" > > clause meaningless at best, disingenuous at worst. > > You have to have a kernel driver for ide, scsi, sata, sas, so why would RAID > be any different?
I don't have a kernel driver for seagate, maxtor, or western digital. Why not? I don't have to load in a new kernel driver when I replace my viewsonic monitor with a dell; I replaced the CRT with a flatscreen and it Just Worked. This is how good design goes. Reduced coupling between subsystems. Printers used to be like this, before the everything-needs-a-custom-driver marketing idiots managed to wedge it back in. > > You can't yank an IDE/SCSI/FC drive out and drop a hardware RAID in. You > > just get hardware acceleration for what is basically a software RAID... > > You can't "yank an IDE/SCSI/FC drive out" and put in any of the others, > without also loading up a different module for the new drive. Sure I can. I can replace an IDE drive with another, larger, faster, IDE drive, and the system Won't Care. > In that sense > RAID is no different. You must "yank" the old module and install the new. RAID is different when the vendors CHOOSE to make it different. Several folks have pointed out that things have changed in the past few years, and if true, then that's a pretty good argument against RAID being intrinsically special. There's a long way between custom drivers being useful and custom drivers being necessary. -- Custom drivers do no good if I'm not running a supported OS. Like linux was. Stewart Stremler -- [email protected] http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
