Skip to main content

We've Moved!

Product Documentation has moved to docs.hitachivantara.com
Hitachi Vantara Knowledge

Capacity design (for HPEC 4D 2P)

Capacity design principles (1) : Physical capacity (for HPEC 4D+2P)

The physical capacity in drives is classified into the following two categories:

  • Valid physical capacity (totalRawCapacity): The amount of space that has been available since the drive was added to the storage pool.

  • Physical capacity of control information (metaDataPhysicalCapacity): The capacity for control information in the total capacity (total physical capacity) of the drives allocated to the storage pool.

For HPEC 4D+2P, the effective physical capacity (CrawDevice[MiB]) is determined by the following calculations:

CrawDevice = floor((Cdevice × 0.9846) - 2048[MiB] - 148932[MiB], 148932[MiB])
    floor(value, reference value)
        : The value is rounded off to a multiple close to the reference
          value.
    Cdevice: Drive capacity[MiB]

GUID-3C4B0B37-1CC1-484C-A71B-9554543253D1-low.png

Capacity design principles (2) : Logical capacity (for HPEC 4D+2P)

The following assumptions are assumed:

  • Each storage node (or drive) in the storage cluster has the same valid physical capacity.

  • The cluster does not have faulty storage nodes or faulty drives.

  • The configuration and the capacity of each storage node satisfy the product specifications.

Logical capacity

The logical capacity (totalCapacity) is the storage pool capacity available to users. To allow Capacity balance to resolve imbalance among the storage controllers, configure logical capacity (totalCapacity) with adequate margin so that planned capacity usage is kept to 80% or below.

You can calculate logical capacity (totalCapacity) by using the formula shown further on in this section.

First, calculate physical capacity used as logical capacity (usablePhysicalCapacity).

Subtract the capacity reserved for configuration changes (893592MiB MiB) and the capacity allocated as the rebuild capacity (CrawDevice×Nrebuild) from the valid physical capacity of each storage node. Out of the remaining capacity, capacity for multiples of 893592MiB is used for physical capacity used as logical capacity (usablePhysicalCapacity).

Physical capacity used as logical capacity [MiB/Node] = RoundDown((CrawNode - 893592-CrawDevice × Nrebuild) / 893592) × 893592
  CrawNode: Valid physical capacity [MiB/Node]
  CrawDevice: Valid physical capacity [MiB/Device]
  Nrebuild: Tolerable number of drive failures set for the storage pool
         (0 if the rebuild capacity policy is set to "Variable")
  RoundDown (numerical value): A numerical value is rounded up to an integer.

Out of every 893592MiB of physical capacity used as logical capacity, 595728 MiB is used for logical capacity.

In addition, each storage controller needs 4200MiB of managed capacity, which is excluded from the logical capacity.

Logical capacity [MiB] = ((CusableNode / 893592 × 595728) - 4200) × Nnode
  CusableNode:  Physical capacity used as logical capacity [MiB/Node]
  Nnode: Number of storage nodes [number of storage nodes / cluster]

As a result of the calculation, if the logical capacity is 0 or less, the logical capacity will not be configured. However, when the calculated logical capacity is 0 or less while Nrebuild is 1 (default value), if the logical capacity can be formed by setting Nrebuild to 0 and then recalculating, the logical capacity will be formed without securing the rebuild capacity.

The capacity that a storage controller can manage has an upper limit. The capacity exceeding the following values cannot be used as logical capacity.

  • Physical capacity used as logical capacity 151017048 [MiB/Node]

  • Logical capacity: 100673832 [MiB/Node]

Reserved physical capacity

Physical reserved capacity (reservedPhysicalCapacity) is the amount of available physical capacity in the storage pool that is not used for logical capacity. The physical reserved capacity is determined by the following calculation:

Reserved physical capacity [MiB] = (CrawNode-CusableNode) × Nnode
  CrawNode: Valid physical capacity [MiB/Node]
  CusableNode: Physical capacity used as logical capacity [MiB/Node]
  Nnode: Number of storage nodes [number of storage nodes / cluster]

GUID-D5C2FF13-3395-42B4-97DF-DA8C70D0B0C6-low.png

Tip

The logicalized physical capacity (usablePhysicalCapacity) shown in the figure above is the amount of effective physical capacity in the storage pool that is available for the logical capacity.

Physical reserved capacity (reservedPhysicalCapacity) is the amount of effective physical capacity in the storage pool that is not used for logical capacity.

Capacity design method (1)-1 : Logical capacity from physical capacity (for HPEC 4D+2P)

The following is assumed:

  • Each storage node (or drive) in the storage cluster has the same valid physical capacity.

  • The storage cluster does not have faulty storage nodes or faulty drives.

  • The configuration and the capacity of each storage node satisfy the product specifications.

The following example procedure assumes that the configuration consists of six storage nodes and each storage node has a capacity of 16 TB (1.6-TB drive × 10) with the rebuild capacity policy set to "Fixed" and the tolerable number of drive failures set to 1.

Procedure

  1. Convert the unit for drive capacity to TiB:

    1.6 × 10004 / 10244 = 1.4552[TiB]
  2. Calculate the valid physical capacity per drive (1.6 TB) by using the expression in section Capacity design principles (1) : Physical capacity (for HPEC 4D+2P).

    floor((1.4552 × 10242 × 0.9846) - 2048[MiB] - 148932[MiB], 148932[MiB])
        = 1340388 [MiB/device]
        floor(value, reference value)
            : The value is rounded off to a multiple close to the reference
              value.
  3. Calculate the valid physical capacity per storage node.

    1340388 × 10 = 13403880[MiB/node]
  4. Calculate the physical capacity used as logical capacity of the storage cluster by using the expression shown in Capacity design principles (2) : Logical capacity (for HPEC 4D+2P).

    (RoundDown((13403880 - 893592 - 1340388) / 893592) × 893592 = 10723104[MiB]
        RoundDown( numerical value )
            : A numerical value is rounded down to an integer.
  5. Calculate the logical capacity of the storage cluster by using the expression shown in Capacity design principles (2) : Logical capacity (for HPEC 4D+2P).

    (10723104 / 893592 × 595728 - 4200) × 6 = 42867216[MiB/Cluster]

Capacity design method (2) : Physical capacity from logical capacity (for HPEC 4D+2P)

The following is assumed:

  • Each storage node (or drive) in the storage cluster has the same valid physical capacity.

  • The storage cluster does not have faulty storage nodes or faulty drives.

  • The configuration and the capacity of each storage node satisfy the product specifications.

In the following example procedure, you calculate the physical capacity required per storage node when a logical capacity of 50 TiB is allocated to six storage nodes (20 device slots). The rebuild capacity policy is set to "Fixed", and the number of tolerable drive failures is set to 1.

Procedure

  1. Since the logical capacity can be increased in 496440-MiB increments (section Capacity design principles (2) : Logical capacity (for HPEC 4D+2P)), the valid physical capacity required in each storage node is:

    RoundUp(((50 × 10242) / 6 + 4200) × (893592 / 595728)) + 893592
        = 14007092[MiB/node]
        RoundUp( numerical value )
            : A numerical value is rounded up to an integer.
  2. Calculate the valid physical capacity of the used drives.

    For example, the valid physical capacity of a drive (1.6 TB) is 1340388 [MiB/device].

    Because the capacity for one drive is allocated as the rebuild capacity, the number of necessary drives can be calculated by using the following expression:

    RoundUp(14007092 / 1340388) + 1 = 12[Drives/Node]
      RoundUp(value): Result of rounding value up to an integer

    Through the preceding calculation, it is found that 50-TiB logical capacity is available by constructing a storage cluster consisting of six storage nodes with twelve 1.6-TB drives.

 

  • Was this article helpful?