* Man-wai Chang to The Door (24000bps):
.... compared to a typical PC hard disk?
Basically the same. Bad sectors are not detected by the computer, they
are detected by the hard drive's controller. Every disk has bad sectors,
it only becomes a problem when there are no spare sectors left to
relocate data. In this case a SMART warning/error is issued to the hard
drive controller.
Mainframe-class computers don't have internal disks, they use storage
systems connected over network-like interfaces like Fibre Channel.
Everything which has to do with disk management is done in the storage
systems which usually use hardware RAID for redundancy and availability,
the available storage space is fully transparent to the computer (it
doesn't "see" hard drive defects). Hardware RAID controllers like HP
Smart Array controllers which are used in standard servers (HP
ProLiants) work similarly, they work fully transparent to the operating
system.
If a SMART warning/error is issued by a disk drive in a mainframe
storage system or on a hardware RAID in a standard server the disk is
flagged as faulty and replaced by the administrator, after which a
rebuild cycle starts (which again is fully transparent to the computer
itself). On a desktop PC, if a SMART warning/error occurs, the user will
get a warning at bootup if the BIOS has proper SMART support. On a Mac,
the user will have to look into the hard disk utility or use other
utilities to check the SMART status.
As not all errors are flagged over SMART, storage systems and hardware
RAID controllers regularly do consistency checks of their hard drives, too.
Benjamin
Benjamin
