Elements of a vdisk set definition

The definition of a vdisk set requires seven attributes and one or more recovery groups:

  1. Vdisk set name: A name for the vdisk set that is unique within the IBM Storage Scale cluster.
  2. Declustered array: The declustered array in the recovery groups where the member vdisk NSDs are located.
  3. RAID code: The vdisk RAID code used for the member vdisk NSDs.
  4. Block size: The file system block size (equivalently, the vdisk track size) of the member vdisk NSDs.
  5. Vdisk NSD size: The usable size of a member vdisk NSD.
  6. NSD usage: The file system usage of a member vdisk NSD.
  7. Storage pool: The file system storage pool of a member vdisk NSD.

    Recovery groups: One or more recovery groups, each of which contributes one member vdisk NSD to the vdisk set.

Defining a vdisk set requires specifying at least one recovery group, the vdisk set name, the RAID code, the block size, and the vdisk set size (from which the member vdisk NSD size is calculated). The declustered array, NSD usage, and the storage pool attributes can be omitted when mmvdisk can determine suitable default values.

Table 1. Vdisk set attributes and their default values
Vdisk set attribute Default value Valid values
vdisk set name must be specified The vdisk set name must be unique within the IBM Storage Scale cluster. The name is limited to the same character set as IBM Storage Scale file system device names and IBM Storage Scale RAID recovery group names.
declustered array name DA1 (if only one DA is present) All recovery groups for which this vdisk set is defined must have a declustered array with this name where this vdisk set's members are created. The expectation is that a vdisk set extends across structurally identical recovery groups where the named declustered array has the same characteristics in each recovery group. If there is only one user declustered array in each recovery group, it is named DA1 and this is the default. If there is more than one user declustered array in a recovery group, there is no default and a declustered array name must be specified.
RAID code must be specified This is the vdisk RAID code for members of the vdisk set. Valid values are: 3WayReplication, 4WayReplication, 4+2P, 4+3P, 8+2P, or 8+3P
block size must be specified This is the file system block size (vdisk track size) for members of the vdisk set. It is constrained by the selected RAID code. Valid values for 3WayReplication and 4WayReplication are 256k, 512k, 1m, or 2m. Valid values for 4+2P and 4+3P are 512k, 1m, 2m, 4m, or 8m. Valid values for 8+2P and 8+3P are 512k, 1m, 2m, 4m, 8m, or 16m.
vdisk set size must be specified The vdisk set size is the desired aggregate size of the vdisk set members in one recovery group. The set size can be specified as a percentage (whole numbers from 1% to 100% using the % suffix) or as a number of bytes (a number, optionally followed by one of the base 2 suffixes K, M, G, or T). If the vdisk set size is given as a percentage, it specifies the raw size to use from the declustered array including RAID code redundancy. If the vdisk set size is given as a number of bytes, it specifies the desired usable size of the vdisk set excluding RAID code redundancy. The vdisk set size is used to calculate the usable size of a single vdisk NSD member of the vdisk set in one recovery group. It is this calculated usable size that becomes part of the vdisk set definition, so if the size of a declustered array should ever change, the size of the individual member vdisk NSDs remains constant.
Note: The resulting actual set size will almost certainly differ from the requested set size, especially if a number of usable bytes is specified, since the IBM Storage Scale RAID internal data structures to accommodate the requested size will depend on the number and size of the pdisks and on the RAID code and block size of the member vdisks.
NSD usage dataAndMetadata This is the IBM Storage Scale file system data usage for the NSD. Valid values are dataAndMetadata, metadataOnly, and dataOnly. The default is dataAndMetadata.
storage pool system If the NSD usage is dataAndMetadata or metadataOnly, the storage pool value must be system and does not need to be specified. If the NSD usage is dataOnly, the storage pool must be specified and the value may not be system.
The following illustration is a visualization of how four vdisk sets might span across four paired recovery groups (two IBM Storage Scale System 3200 recovery group pairs), and how they are used to form three file systems:
                             
                                 File System           File System   File System
                                     FSA                   FSB           FSC
                          /-----------------------\    /---------\   /---------\
                           Vdisk Set     Vdisk Set      Vdisk Set     Vdisk Set
                             VSA1          VSA2            VSB           VSC
                          /--------\     /--------\     /-------\     /-------\
 Recovery Group ESS01L {  VSA1 vdisk  |  VSA2 vdisk  |  VSB vdisk  |  VSC vdisk  }
 Recovery Group ESS01R {  VSA1 vdisk  |  VSA2 vdisk  |  VSB vdisk  |  VSC vdisk  }
 Recovery Group ESS02L {  VSA1 vdisk  |  VSA2 vdisk  |  VSB vdisk  |  VSC vdisk  }
 Recovery Group ESS02R {  VSA1 vdisk  |  VSA2 vdisk  |  VSB vdisk  |  VSC vdisk  }
                          \--------/     \--------/     \-------/     \-------/
                           Vdisk Set      Vdisk Set     Vdisk Set     Vdisk Set
                             VSA1           VSA2           VSB           VSC
                          \-----------------------/    \---------/   \---------/
                                File System            File System   File System
                                    FSA                    FSB           FSC

There are four structurally identical recovery groups built from two IBM Storage Scale System 3200 building blocks.

There are three file systems all uniformly striped across the recovery groups using four vdisk sets. An identical member of each of the four sets vdisk sets is present in each recovery group.

File system FSA uses two vdisk sets, VSA1 and VSA2.

File system FSB uses one vdisk set, VSB.

File system FSC uses one vdisk set, VSC.

For the high-level purpose of visualizing how four vdisk sets might span four recovery groups to construct three file systems, details such as the RAID code and block size for the four vdisk sets are omitted. It is enough to know that within a vdisk set, the details are the same for each member vdisk NSD.

This illustration also shows how the concept of vdisk sets helps to accomplish the IBM Storage Scale RAID best practice of constructing file systems from identical vdisk NSDs equally distributed across all recovery groups.