When any hard disk detects a problem with a particular sector on the surface of the
drive, it will remap that sector and mark it as "bad" so that it will not be
written to again in the future. This ensures that repeat problems with this bad spot will
not recur. This process is called remapping and spare sectoring and is described
in this section on hard disk formatting.
Some drives go a step beyond ordinary remapping. Instead of just remapping the sector
where an error is encountered, or where a number of retries were necessary to get the data
from the disk, the controller remaps a zone of sectors around the defective location. The
logic behind this feature is that if an area of the disk is damaged badly enough to create
a bad sector, the problem might not be limited to just that sector, even if the errors are
only showing up there right now. After all, a single sector on a hard disk is very
small; it stands to reason that there is a much higher chance that a sector near the
defective one will go bad than a random sector somewhere else on the disk.
The "buffer area" that is remapped can include sectors before and after the
bad sector, as well as adjacent tracks, thus covering a two-dimensional space centered on
the bad disk location. This process occurs transparently to the user. It makes use of
spare sectors allocated on the drive for the purpose of remapping.
Western Digital calls this feature defect margining. As with the wear leveling feature, defect margining does
not appear to have been universally adopted by other hard disk manufacturers. Unlike
wear leveling, there is not to my knowledge any major drawback to this feature, and it
seems in theory to be a very good idea. (I suppose it might require a few extra spare
sectors to be placed on the drive, but capacity is very cheap these days.) I do not know
why the other drive makers are not also doing this--maybe they are doing it but
just not publicizing it.
Next: Redundant Arrays of Inexpensive Disks (RAID)
