DELL PowerVault MD3200i/MD3220i Technical Guide Book
each physical disk stores that physical disk's state and status, the worldwide name of its disk group,
the virtual disks it contains and the definitions for those virtual disks.
Storing metadata in this nonvolatile region provides the highest availability and enables easier
reconfigurations and migrations. As system configuration data resides on every configured physical
disk, controllers and/or multiple physical disks can be removed or swapped without losing the system
configuration. Physical disks can be relocated within the storage system to improve channel
utilization/protection or even migrated as a complete disk group into another storage system. In
instances, all configuration metadata and user data remains intact on the physical disks.
Global Hot Spares
If a physical disk fails in the MD3200i or MD3220i storage array, the controller uses redundancy data to
reconstruct the data onto a hot spare physical disk. The hot spare is automatically substituted for the
failed physical disk without requiring user intervention. Once the failed drive is replaced, the data is
then automatically copied back to the new drive. The user then has a choice of allowing the hot spare
to return to its role as a hot spare drive or become a permanent member of a disk group, therefore,
eliminating the copy back process. The replacement drive, in this case, would then be in an
unassigned state. This feature may be appealing as it reduces the degraded mode time by avoiding the
copy back process. The MD3200i and MD3220i storage arrays both support an unlimited number of
global hot spare physical disks, and each can be a spare for any like-technology disk in the array (i.e.
SAS for SAS, near-line SAS for near-line SAS and SSD for SSD). Generally speaking, the physical disks
that are assigned as a hot spares should have a capacity that is equal to or greater than the capacity of
the largest physical disk in the storage array.
Storage Partitioning
The MD3200i series of arrays Storage Partitioning feature enables a single
storage array to be logically partitioned and function as up to 32 virtual
arrays. A storage partition is a logical entity consisting of one or more
virtual disks that are accessed by a single host or shared among a collection
of hosts that are part of a host group. A storage partition is created when
one or more virtual disks are mapped to an individual host or host group.
This virtual disk-to-LUN mapping allows you to define what host or host
group will have access to a particular virtual disk in your storage system.
Hosts and host groups can only access data through assigned virtual disk-to-
LUN mappings. Partition access is maintained at the controller level,
ensuring complete data integrity in multi-host, multi-OS environments.
The virtual disk-to-LUN mapping implementation creates valuable flexibility
for the storage administrator as any available virtual disk can be mapped to
any attached server. So, while the individual servers see a virtual array that
consists of only their LUNs/virtual disks, the virtual disks can be intermixed
throughout the storage system within one or more disk groups.
The system's logical partitioning combined with its configuration capabilities
enables administrators to choose from a range of virtual disks with different
characteristics to meet a server's exact needs for a given application. A
given host can concurrently have virtual disks with different capacities,
different RAID levels, and different physical disk types. This flexibility enables hosts with different
capacity, performance or data protection demands to effectively share a single storage array.
Host B
Host A
Logical
Logical
partition
partition
A
B
L U N
L U N
L U N
L U N
L U N
u n m a p p e d
v i r t u a l d i s k
30
Печатать страницу