REFORMAT of SCSI disks, useful or waste of time?
Gerry_Houlder at notes.seagate.com
Fri Mar 10 05:42:23 PST 1995
Here are my answers to Ken's questions:
Question: When I (OEM user of 1000 disk/year) reformats a SCSI disk, does
the on board SCSI controller really attempt to reformat the platters?
What am I not understanding about reformat of SCSI disk?
Answer: Some implementations take only a few seconds to do the Format command.
This implementation is just initializing its logical to physical mapping tables
to take care of defective blocks and set up the block size requested by the
customer. This doesn't actually rewrite any of the tracks in the user area.
Other implementations take many minutes (or up to 2 hours for the largest
capacities now available) to do a Format. This implementation is re-writing the
tracks in the user area with default data patterns and new headers (if the
implementation uses headers at all). Consult your local product specification
to see what your drive does.
Question: Currently we reformat all drives that are shipped to us by the
manufacturer. Does this help to weed out and blocks that could have gone
bad as a result of SHIPMENT or are we wasting our time?
Answer: I wouldn't trust the format process (either of the processes described
above) to detect new bad spots. A pack scan utility that writes and reads back
a number of different data patterns with ECC correction disabled would be much
more effective than doing a Format. If the drive does re-write the tracks in
the user data area, a format would help by compensating for heads that are
mis-aligned due to shock but modern drives have a lot of built in capability
for this anyway (with embedded servos, thermal calibration, etc. that
compensate for individual head misalignments). A Format might be useful on
cheaper drives that have less range in their head compensation schemes, but in
general you are wasting your time.
More information about the T10