IBM DS8900F, Introduction And Planning Manual

Page: 72 / 172

slots in the High Performance Flash Enclosures Gen2 pair. If two drive sets are ordered (32 drives),
one filler set is require to fill the remaining 16 slots. Each drive slot in a High Performance Flash
Enclosures Gen2 must have either a flash drive or a filler.
Physical and effective capacity
Use the following information to calculate the physical and effective capacity of a storage system.
To calculate the total physical capacity of a storage system, multiply each drive-set feature by its total
physical capacity and sum the values. For the standard drive enclosures, a full drive-set feature consists
of 16 identical disk drives with the same drive type, capacity, and speed. For High Performance Flash
Enclosures Gen2, there are 16 identical flash drives.
The logical configuration of your storage affects the effective capacity of the drive set.
Specifically, effective capacities vary depending on the following configurations:
RAID type and spares
Drives in the DS8000 must be configured as RAID 5, RAID 6, or RAID 10 arrays before they can be
used, and then spare drives are assigned. RAID 10 can offer better performance for selected
applications, in particular, high random, write content applications in the open systems environment.
RAID 6 increases data protection by adding an extra layer of parity over the RAID 5 implementation.
Data format
Arrays are logically configured and formatted as fixed block (FB) or count key data (CKD) ranks. Data
that is accessed by open systems hosts or Linux on IBM Z that support Fibre Channel protocol must
be logically configured as FB. Data that is accessed by IBM Z hosts with z/OS or z/VM must be
configured as CKD. Each RAID array is divided into equal-sized segments that are known as extents.
The storage administrator has the choice to create extent pools of different extent sizes. The
supported extent sizes for FB volumes are 1 GB or 16 MB and for CKD volumes it is one 3390 Mod1,
which is 1113 cylinders or 21 cylinders. An extent pool cannot have a mix of different extent sizes.
On prior models of DS8000 series, a fixed area on each rank was assigned to be used for volume
metadata, which reduced the amount of space available for use by volumes. In the DS8900F family, there
is no fixed area for volume metadata, and this capacity is added to the space available for use. The
metadata is allocated in the storage pool when volumes are created and is referred to as the pool
overhead.
The amount of space that can be allocated by volumes is variable and depends on both the number of
volumes and the logical capacity of these volumes. If thin provisioning is used, then the metadata is
allocated for the entire volume when the volume is created, and not when extents are used, so over-
provisioned environments have more metadata.
Metadata is allocated in units that are called metadata extents, which are 16 MB for FB data and 21
cylinders for CKD data. There are 64 metadata extents in each user extent for FB and 53 for CKD. The
metadata space usage is as follows:
• Each volume takes one metadata extent.
• Ten extents (or part thereof) for the volume take one metadata extent.
For example, both a 3390-3 and a 3390-9 volume each take two metadata extents and a 128 GB FB
volume takes 14 metadata extents.
Note: In a multiple tier pool volume, metadata is allocated on the upper tiers to provide maximum
performance. A pool with 10% Flash/SSD or greater would have all of the volume metadata on this tier.
A simple way of estimating the maximum space that might be used by volume metadata is to use the
following calculations:
FB Pool Overhead = (#volumes*2 + total volume extents / 10)/64 - rounded up
to the nearest integer
CKD Pool Overhead = (#volumes*2 + total volume extents / 10)/53 - rounded up
to the nearest integer
58 IBM DS8900F: DS8900F Introduction and Planning Guide