EMC Symmetrix DMX-3, Product Manual

Page: 203 / 326

Symmetrix DMX RAID 5 203
Data Integrity, Availability, and Protection
If a physical device reports too many errors (even recoverable errors),
or fails completely, that device is taken offline by the Symmetrix
system. This condition causes the Symmetrix DMX system to place a
remote service call to the Customer Support Center.
The Product Support Engineer (PSE) at the support center can
determine if a disk drive should be identified for replacement and
dispatches a Customer Engineer (CE) who starts the Symmetrix
nondisruptive interactive drive replacement process. The logical
devices on the disk being replaced are placed in a not-ready state and
the associated ranks will begin operating in regeneration mode. After
the new drive is in place, the regeneration process begins.
RAID 5 recovery with a
dynamic spare or
permanent spare
If the Symmetrix DMX system has an available dynamic spare or
permanent spare, it copies the data from the failing device to the
spare before the device fails, and reconstructs the data if necessary
from the other surviving members of the RAID 5 device. After a spare
is activated, it behaves as if it is a full fledged member of the RAID 5
device.
Note: “Dynamic sparing with RAID 5 volumes” on page 219 contains
information on RAID 5 regeneration with dynamic sparing and “Permanent
sparing” on page 213 for RAID 5 regeneration with permanent sparing.
Write requests during
regeneration
As write requests are made during regeneration, the new data is
written to the RAID 5 device consisting of the surviving members
and the dynamic spare or permanent spare
if one is available
.
Read requests during
regeneration
Read requests for data in global memory (read hits) are serviced
immediately from global memory. During regeneration, when read
requests not found in global memory (read misses) occur, the
required data is automatically regenerated by reading surviving
members and using an XOR calculation, and is delivered to the
spares pool.
Fast Copy Back after
replacing a disk using
dynamic sparing
After the failed disk is replaced, data is rapidly copied back from the
the dynamic spare
to the newly installed drive. This is a fast process
of simply copying track by track. After copy back is completed, the
original spare drive is returned to the dynamic spare pool.