Table of Contents
Advertisement
Estimating compression savings
Compressible data can be identified and expected compression ratios can be
estimated even before using compression.
On an XIV system supporting compression, the compression ratio for all
uncompressed volumes in the system is continuously estimated, even before
enabling compression. The decision to use compression can be based on the
expected storage savings of the compressed data and the expected effect on
performance (throughput and latency) of the compression processing overhead.
Information on compression usage can also be monitored using the XIV GUI to
determine the potential savings to your storage capacity when uncompressed
volumes are compressed. You can view the total percentage and total size of
capacity savings when compression is used on the system. Compression savings
across individual domains, pools and volumes can also be monitored. These
compression values can be used to determine which volumes have achieved the
highest compression savings. See the IBM XIV Storage System Management Tools
User Guide for more information on monitoring and using compression.
Note: Keep in mind that the expected storage savings can vary from 5% higher or
lower than the actual compression ratio. A negative estimated compression ratio
can be due to metadata that consumes storage space on a volume, even when a 0%
estimated compression ratio is received from data that cannot be compressed.
Effective capacity
Effective capacity is the amount of storage that is allocated to applications.
Using thin-provisioned storage architectures, the effective capacity is virtually
larger than the array usable capacity. This is made possible by over-committing
capacity, or by compressing the served data. Compression is the preferred method
to apply thin-provisioning to usable capacity, since the compression ratio is highly
predictable.
Hard capacity denotes usable, non-compressed capacity, whereas soft capacity
denotes the nominal capacity that is assigned to volumes, and reported to any
hosts mapped to those volumes. Thin provisioning denotes committing more soft
capacity than hard capacity. Soft capacity is assigned at a pool level. In the case of
compression, thin-provisioning is obvious: a compressed volume will forever use
less hard capacity than soft capacity.
In XIV terms, the effective capacity is allocated out of the system soft capacity, and
is the sum of the sizes of all the allocated volumes.
In an XIV System Gen3 Turbo Compression model 314, the maximal,
non-compressed hard capacity supported in a single XIV frame is 485 TB (15
modules, 6 TB drives). However, an XIV frame can effectively accommodate up to
2 PB of real written data - when the data is compressed. With very high
compression rates, filling a system up to 2 PB of soft capacity may not require a lot
of usable capacity.
The maximum soft capacity that can be allocated to volumes in XIV is 2 PB.
Considering the compression ratio for typical data profiles on XIV systems, the
effective soft capacity leveraged within a 15-module frame will range from 1 PB to
2 PB. To maximize the utilization of XIV hard and soft capacity with compressed
7
Chapter 3. Planning for compression
Advertisement
Table of Contents
