HP Logical Server Management Best Practices

13
Storage specification
The specification of storage for a logical server can be done in one of several ways, based on the administrator
needs and preferences.
The system administrator can specify the storage LUN information for each logical server individually at the
time of creation. Logical server storage pool entries are created as required, and the administrator can
choose if they are populated into the storage pool for the portability group or only accessible through the
logical server definition.
The system administrator can specify use of storage pool entries at the time of logical server creation (with
storage pools being created and maintained by either the system or storage administrator, using the GUI
interface or command-line interface).
The storage required for a logical server can be captured when the system administrator imports an existing
blade or virtual machine to create a logical server.
Matrix infrastructure orchestration can automate the creation of the logical servers and auto-generate the
storage pool entries necessary for the service template (or match to existing manually defined storage pool
entries)
Storage specification at the time of create provides a fast path for those system administrators wanting to consider the
unique needs of that one logical server. That storage can also be placed in the storage pool, enabling a
comprehensive view of all storage defined for logical servers within that portability group. It is a best practice to put
entries into the storage pools, as it enables both the comprehensive view of the storage and possible re-use of the
entry if the logical server using it is deleted. It is necessary to define storage pool entries in advance of logical server
definition to take advantage of the HP SPM integration with the ability to select from among storage catalog matching
candidates. If the storage is being defined during the definition of the logical server, there is no storage catalog
integration.
The use of predefined storage pool entries provides the most flexibility. The storage pool can be pre-populated with
storage and subsequent logical server creations consume pool entries. Matrix OE includes full support for integration
with the HP Storage Provisioning Manager. This enables the storage administrator to create a catalog of imported
storage arrays and pools, and define storage templates. If desired, pre-provisioned storage volumes can be
imported and made available for use by the server administrator, or on-demand provisioning can create volumes as
they are required to fulfill Matrix storage requests. The server administrator can populate the storage pool by
requesting matching volumes from the storage catalog and choosing an appropriate candidate (without needing to
engage the storage administrator for each individual request). It is a best practice to use the HP Storage Provisioning
Manager integration for faster fulfillment of requests (automated matching against a pre-populated pool or on-demand
provisioning) and automatic population of storage details (such as controller port WWNs and LUN number). There is
also support for environments in which the storage administrator does not want to pre-populate the storage catalog or
enable on-demand provisioning.
When an existing Virtual Connect blade is imported to become managed as a logical server, the storage information
from the Virtual Connect profile will be used to construct a storage entry. That entry will contain information about
server HBA/CNA initiator WWNs, fabrics, and the FC SAN boot volume. The OS type will be assumed to be
Windows. The administrator can modify the logical server, adjusting the storage information as appropriate (e.g.,
changing the host mode, specifying a RAID level, adding data volume information). The administrator may choose to
validate the storage information provided against the actual storage definition on the P6000/EVA or MSA array. The
administrator can also specify that the storage entry is to be inserted into the storage pool.
There are a variety of ways logical server storage pool entries can be used:
One entry may contain all storage needed for a given logical server (boot and data volumes). All volumes
may be present initially, or data volumes may be added after the OS is installed. Later sections of this white
paper provide more details.
Separate entries may be used, with a single logical server consuming multiple storage pool entries. This is
often done when one entry represents a shared data volume (perhaps to be shared by four members of a
cluster solution), and separate entries represent the boot volumes for each of those four logical servers.
Multiple entries can also be used to separate boot and data for non-sharing environments, but there are
implications for multi-pathing for environments that do not support multi-initiator NPIV, as discussed later in
this white paper. Use of multiple storage pool entries also consumes additional initiator WWNs.