You can migrate data on partitioned IBM® System Storage® DS5000 and IBM DS4000® systems.
You can enable the
SAN Volume Controller to
be introduced to an existing SAN environment, so that you have the
option of using image mode LUNs to import the existing data into the
virtualization environment without requiring a backup and restore
cycle. Each partition can only access a unique set of HBA ports, as
defined by the worldwide port names (WWPNs). For a single host to
access multiple partitions, unique host fibre ports (WWPNs) must be
assigned to each partition. All LUNs within a partition are identified
to the assigned host fibre ports (no subpartition LUN mapping).
Host A is mapped to LUN 0, 1, 2 in Partition 0.
Host B is mapped to LUN 0, 1, 2, 3, 4, 5 in Partition 1.
Host C is mapped to LUN 0, 1, 2 in Partition 2.
To allow Host A to access the LUNs in partition B, you must
remove one of the HBAs (for example, A1) from the access list for
partition 0 and add it to partition 1. A1 cannot be on the access
list for more than one partition.
To add a SAN Volume Controller into
this configuration without backup and restore cycles requires a set
of unique SAN Volume Controller HBA
port WWPNs for each partition. This allows the IBM System Storage DS5000 or IBM DS4000 system
to make the LUNs known to the SAN Volume Controller,
which then configures these LUNs as image-mode LUNs and identifies
them to the required hosts. This violates a requirement that all SAN Volume Controller nodes
must be able to see all back-end storage. For example, to fix this
problem for an IBM DS4000 system,
change the configuration to allow more than 32 LUNs in one storage
partition, so that you can move all the LUNs from all the other partitions
into one partition and map to the SAN Volume Controller cluster.
Scenario: the SAN Volume Controller nodes
cannot see all back-end storage
The IBM DS4000 series
has eight partitions with 30 LUNs in each.
Perform the following
steps to allow the
SAN Volume Controller nodes
to see all back-end storage:
- Change the mappings for the first four partitions on the IBM DS4000 system
such that each partition is mapped to one port on each node. This
maintains redundancy across the cluster.
- Create a new partition on the system that is mapped to all four
ports on all the nodes.
- Gradually migrate the data into the managed disks (MDisks) in
the target partition. As storage is freed from the source partitions,
it can be reused as new storage in the target partition. As partitions
are deleted, new partitions that must be migrated can be mapped and
migrated in the same way. The host side data access and integrity
is maintained throughout this process.