HP Matrix Operating Environment Automated Storage Provisioning: "Static"SAN volume automation via multi-initiator NPIV

Insight Orchestration – Storage Automation via NPIV

Figure 2: Deploy SAN storage from a storage pool

Figure 2 modifies the process by introducing the concept of a storage pool. Each entry in the storage

pool can contain one or more volumes. In addition, the storage pool (in its simplest case) also

contains a pair of server HBA WWNs. The storage pool entry forms the nexus of one or more SAN

volumes and the server identity that will be used to access SAN volumes. With storage pools, the

Server administrator is able transition from a “one server at a time” discussion with the SAN

administrator to a discussion regarding the number of servers that will be provisioned in the next six

months, their storage requirements, and how they will be used (i.e. compute intensive, sequential read

intensive, block read intensive, …).

With this information, the SAN administrator can pre-provision, on a less frequent basis, multiple

storage pool entries for the Server administrator to consume during the six month period. The storage

pool forms the boundary within which the Server administrator my flexibly create, deploy, delete, and

re-create physical servers.

This leads to both a process and management experience which is analogous to the one followed

when managing virtual machines. It is accomplished without violating the existing data center

management or process protocols. Each administrator continues to use the tools and processes that

are unique to his management domain.

Once the SAN storage provisioning is complete, no further changes are needed when the Server

administrator assigns on or more of the storage pool entries to a particular server. The SAN fabric

definition remains stable and no changes need to be made to the presenting disk array.

Using the idea of a pre-provisioned storage pool entry as a building block, we can now extend the

concept slightly to include the ability to separate storage pool entries into three broad categories:

1. Boot storage

2. Private data storage

3. Shared data storage

As described above, each storage pool entry minimally contains a pair of server HBA WWNs. The

server HBA WWNs effectively become a “key” which, when given to a particular server, allows that

server to access the pre-provisioned storage resource defined by the storage pool entry. Because we

have separated the storage pool entries into three categories, we are able to sequence the server’s

access to each of the storage categories in time. This is an important point when attempting to

automate the OS deployment process for a server because it is not always possible to

programmatically guide the OS deployment tool to select the desired boot disk while excluding the

visible private or shared data disks.

NPIV (N_Port Id Virtualization)

The key technology used to grant a server visibility to the different types of statically provisioned

storage is known as N_Port Id Virtualization or NPIV for short. NPIV is an extension to the

FibreChannel (FC) specification

which facilitates the sharing of a single physical N_Port across

multiple N_Port IDs. The result is that multiple initiators (each assigned its own N_Port ID) are enabled

to share the same physical port. As the server boots, a fabric login sequence (FLOGI) is performed

using the server HBA’s base port WWN. At the completion of the sequence, the fabric has assigned

one N_Port ID to the base port WWN and the server is granted access to its storage resources based

upon this WWN->N_Port ID mapping.