|
NAS and URE
Modern NASes often use the RAID5 technology to provide fault tolerance,
and therefore protect a user against data loss in case of a disk failure.
In 2007, Robin Harris published an article
which shows that RAID5 cannot be considered a reliable storage technology since the chance to encounter
unrecoverable read error (URE) on a disk after one of the disks has already failed is quite high.
In fact, this article should be taken with a grain of salt (it has some inconsistencies),
but we should not fully reject the idea that this may happen in real life.
Saying this we do not mean a full failure of two disks, but instead the situation when one member disk has died,
while the other has a bad sector(s), and therefore the rebuild cannot be done successfully.
Since a NAS is often used as an archive implying that a user mostly writes data rather than reads,
the phenomenon of URE may really occur. To avoid this and not “oversleep” a moment of replacing the disk you need
to configure your NAS so that the disks are read automatically from the beginning to the end on a regular basis.
In various NASes this may be called "scrub", "read patrol", or simply "disk test".
This allows you to identify bad sectors in time.
|
|