Dell EMC PowerMax: Data Reduction

Inline compression and deduplication

January 2022

H17072.4

White Paper

Abstract
This document describes Dell EMC PowerMax Data Reduction, a
storage-efficiency feature that combines inline compression and inline
deduplication. Using both storage efficiency features together
enhances capacity savings while maintaining great performance and
reliability.

Dell Technologies
 Copyright

The information in this publication is provided as is. Dell Inc. makes no representations or warranties of any kind with respect
to the information in this publication, and specifically disclaims implied warranties of merchantability or fitness for a particular
purpose.
Use, copying, and distribution of any software described in this publication requires an applicable software license.
Copyright © 2018-2022 Dell Inc. or its subsidiaries. All Rights Reserved. Dell Technologies, Dell, EMC, Dell EMC and other
trademarks are trademarks of Dell Inc. or its subsidiaries. Intel, the Intel logo, the Intel Inside logo and Xeon are trademarks
of Intel Corporation in the U.S. and/or other countries. Other trademarks may be trademarks of their respective owners.
Published in the USA January 2022 H17072.4.
Dell Inc. believes the information in this document is accurate as of its publication date. The information is subject to change
without notice.

2 Dell EMC PowerMax: Data Reduction

Inline compression and deduplication
 Contents

Contents
Executive summary.......................................................................................................................4

Data reduction overview ...............................................................................................................4

Configuration details ....................................................................................................................6

Adaptive Compression Engine.....................................................................................................6

Deduplication ................................................................................................................................8

Capacity usage ............................................................................................................................10

Data Reduction Settings .............................................................................................................11

System Efficiency........................................................................................................................11

Supported data services .............................................................................................................17

Conclusion...................................................................................................................................18

Appendix: Technical support and resources ............................................................................19

Dell EMC PowerMax: Data Reduction 3

Inline compression and deduplication
 Executive summary

Executive summary

Overview Data reduction with the Dell EMC PowerMax system delivers a boost to system efficiency
by combining inline compression, inline deduplication and pattern detection. Paring these
techniques of capacity savings creates a system where users can achieve great capacity
savings on reducible data. Data Reduction not only compresses data it also eliminates
redundant copies of compressed data and delivers great performance. The contents of
this technical white paper are intended to inform the reader how Data Reduction functions
within the Dell EMC PowerMax systems.

Revisions Date Description

May 2018 Initial release

September 2020 Updated for PowerMaxOS Q3 2020 release

February 2021 Updated for PowerMaxOS Q1 2021 release

January 2022 Updated template

We value your Dell Technologies and the authors of this document welcome your feedback on this
feedback document. Contact the Dell Technologies team by email.

Author: Robert Tasker

Note: For links to other documentation for this topic, see the PowerMax Info Hub.

Data reduction overview

Introduction In PowerMax data storage systems data reduction combines the Adaptive Compression
Engine (ACE) and inline deduplication to provide a highly performing space efficient
platform. Data reduction allows users to write more host data than the total amount of
usable capacity available. Compression and deduplication are two different functions that
work together. Compression reduces the size of datasets and deduplication identifies
identical datasets and stores a single instance. Performing these functions in parallel
allows the system to be capacity efficient and deliver optimal performance.

Adaptive The Adaptive Compression Engine (ACE) is the combination of multiple components to
Compression deliver the performance expected from an all flash storage system and maintain data
Engine overview storage efficiency. Incoming data is compressed inline using hardware-based
compression with software compression in place and used as needed. Intelligent
algorithms learn from the incoming workload to dynamically create a customized backend
catering to the incoming workload. The Adaptive Compression Engine changes the
backend compression pool layout as needed to ensure the system operates at optimal
levels for both performance and space efficiency. Using internal statistics, algorithms
identify the busiest data in the system allowing it to skip the compression process. The
result minimizes decompression overhead on the system for data that is accessed the

4 Dell EMC PowerMax: Data Reduction

Inline compression and deduplication
 Data reduction overview

most. Working together, these functions allow data reduction to deliver great performance
and efficiently manage back-end capacity usage.

Deduplication Deduplication (dedupe) is a capacity-savings method that identifies identical copies of

overview data and stores a single instance. There are a few facets of deduplication that are needed
for it to provide efficient capacity savings.

• Hash ID: The Hash ID is a unique identifier for incoming data that is used to
determine if a dedupe relationship is needed. The system uses a SHA-256
algorithm to generate the Hash ID.
• Hash ID Table: Hash Tables are an allocation of system memory distributed
between the system directors. These tables are a catalog of the Hash IDs used by
the dedupe process to determine if a dedupe relationship is needed or if the data
can be stored on disk.
• Dedupe Management Object (DMO): The DMO manages the pointers for
deduped data between front-end devices and the data stored on disk. This also
manages what Hash Table the Hash IDs are stored in when dedupe relationships
exist.

Terminology Data Reduction: The use of compression, deduplication and pattern detection to reduce
capacity usage and the cost of physical storage. In systems prior to the release of Dell
EMC PowerMax, data reduction is compression only.

Data Reduction Reservation: A system Cache reservation used to determine the

maximum savings that can be achieved when data reduction is enabled. This reservation
acts as a system guardrail for management of resources such as disk capacity and cache
in order to maintain a balance between performance and efficient capacity usage.

Storage Group Compression Ratio: The compression ratio displayed for allocations
related to a specific storage group. This value may be greater than or less than the
system or storage resource pool data reduction ratio displayed in management application
displays.

Compressibility: The maximum compression ratio that may be achieved for either a
storage group or a device. This value may be presented as a higher value than current
savings due to the design activity-based compression (see Activity Based Compression).

Data Reduction Ready: The state of the system when the default storage resource pool
(SRP) can store compressed data. For a system to be able to compress data it must have
one compression I/O module per director, have compression enabled and have a system
data reduction reservation ratio set.

Data Reduction Capable: A system installed with at least the Q1 2018 PowerMaxOS
where the data reduction reservation applied to the system IMPL is 1.0:1.

Compression Pool: The collection of data devices configured within the physical disks
where the track size is the same. For example, the 64KB pool is made up of data devices
where all the device’s tracks are 64KB in size.

Dell EMC PowerMax: Data Reduction 5

Inline compression and deduplication
 Configuration details

Terabytes Usable (TBu): The backend usable storage capacity in the absence of
compression referring to the amount of physical storage in the system.

Example: 50 TBu is 50 Terabytes of usable physical storage.

Terabytes Effective (TBe): The front-end effective storage capacity in the presence of
data reduction. This represents the potential maximum amount of host or application data
that can be written to the array.

Example: 50 TBu of physical storage with a data reduction reservation of 3:1 translates to
150 TBe. The total TBe value can be achieved assuming the data consuming capacity on
the array is reducible at a level equal to or greater than the data reduction reservation set
in the system.

Configuration details
PowerMaxOS is supported on PowerMax and VMAX All Flash data storage arrays. There
are a few different scenarios for the two storage arrays, as shown in the following table.

Table 1. Configuration details by storage platform

PowerMax VMAX All Flash

Adaptive Compression Engine (ACE) Yes Yes

Extended Data Compression (EDC) Yes No

Inline Deduplication Yes No

Data Reduction I/O Module Yes No

Compression I/O Module No Yes

DEFLATE Compression Algorithm Yes No

LZS Compression Algorithm No Yes

FBA Storage Resource Pool (SRP) Yes Yes

CKD Storage Resource Pool (SRP) No No

Adaptive Compression Engine

Introduction The Adaptive Compression Engine (ACE) is the combination of multiple core components
that work together to achieve maximum system efficiency and deliver optimized
performance. These core components are:

• Hardware acceleration
• Optimized data placement
• Activity Based Compression
• Fine grain data packing
• Extended Data Compression

6 Dell EMC PowerMax: Data Reduction

Inline compression and deduplication
 Adaptive Compression Engine

Hardware Each system is equipped with data reduction hardware that handles the actual
acceleration compressing and decompressing of data. For PowerMax systems where deduplication
applies the data reduction hardware also generates a unique hash ID needed for the
deduplication process. The arrays are configured with one module per director which
equates to 2 for each engine. The use of the modules reduces data reduction processing
overhead. As a secondary function, software compression is automatically applied in the
event of a fault or failure with one or more of the data reduction modules.

Optimized data To maximize data reduction efficiency the system needs to accommodate multiple sizes
placement of compressed data. To support a variety of compression sizes, multiple compression
pools are used to create an optimal backend. Optimized data placement is the function
responsible for dynamically changing the compression pools as needed. This alters the
backend by creating various compression pools that cater to the incoming data. The
result is an evolving layout of compression pools that dynamically change to match the
reducibility of data sent to the system.

Compression pools are identified by labels which represent the track size for the data
devices within the pool. For example, the 128KB pool is made up of data devices where
the tracks are all 128KB in size. The 8KB pool is made up of data devices where the
tracks are all 8KB in size. In comparison the capacity of the data devices between the
pools is the same, however, the 8KB pool has 16 times the number of tracks. Here is a
complete list of possible compression pools; 8KB, 16KB, 24KB, 32KB, 40KB, 48KB,
56KB, 64KB, 72KB, 80KB, 88KB, 96KB, 104KB, 112KB, and 128KB. Due to the dynamic
design each compression-enabled system may have a different combination of
compression pools that reduced data populates.

Activity Based Activity Based Compression (ABC) aims to prevent constant compression and
Compression decompression of data that is accessed frequently. This function allows the busiest data
to avoid being compressed. It differentiates busy data from less busy data and accounts
for up to 20% of the allocations in the SRP. Allowing the busiest allocations to skip
compression is a benefit to the system as well as to end users. This ensures optimal
performance and reduced overhead that can result from constantly decompressing
frequently accessed data. The mechanism used to determine the busiest data does not
add additional load on the system. ABC leverages statistics collected from incoming I/O to
the frontend devices to determine what data sets are the busiest and best candidates to
skip compression. It allows the system to maintain balance the system resources
providing an optimal environment for both data reduction savings and performance.

Fine grain data The Adaptive Compression Engine uses data reduction hardware to process incoming
packing data which is divided into four sections. Each section is compressed in parallel which
maximizes the efficiency of the data reduction module. The sum of the four compressed
sections is the final compressed size and determines where the data is to be stored. In
PowerMax systems where deduplication applies a unique hash id is applied to the
compressed data set. This process includes pattern detection, a non-zero allocate
function. Pattern detection prevents the allocation of any of the four sections that contain
all zeros. This behavior results in an efficient data reduction process that has minimal cost
to performance.

Dell EMC PowerMax: Data Reduction 7

Inline compression and deduplication
 Deduplication

Another benefit of dividing the extents into four sections comes when there are partial
read or write operations. In this case only the sections that contain the requested data are
processed. This means each section can be handled independently.

The efficiency of data compression is measured in terms of the compression ratio. This is
the ratio between the original size of the data and its size after being compressed. For
example, a 128KB dataset is compressed to 64KB, resulting in a compression ratio of 2:1.

Extended Data PowerMax systems include an additional function that will compress already compressed
Compression data to gain further capacity savings. The goal of extended data compression (EDC) is to
apply additional compression savings to already compressed data. This is accomplished
by identifying data that has not been accessed for a set amount of time. The factors that
make data a candidate for EDC are the following:
• The data belongs to a data reduction enabled storage group.
• The data has not been accessed for 30 days.
• The data is not already compressed by EDC.

Data that qualifies for EDC is compressed using the Def9_128_SW algorithm and moved
to the appropriate compression pool. This is an automated background process within the
system. Additional savings are included in the storage group level achieved compression
ratio. EDC is only available with PowerMax storage arrays.

Deduplication

Introduction Deduplication (dedupe) is the process of reducing redundant copies of data that consume
storage capacity. The redundant copies are replaced with pointers. The pointers provide
the access for the subsequent requests of that shared data by multiple sources. In
PowerMax systems, dedupe is accomplished through a series of functions and
components including hardware acceleration, dedupe algorithm, hash table and dedupe
management object (DMO).

Hardware Dedupe is an inline process that uses the same data reduction hardware as compression.
acceleration All data reduction enabled incoming data is passed through data reduction hardware. In a
single pass, the data reduction hardware handles compression, pattern detection and
generates a hash ID for deduplication. This produces compressed data with a unique
hash ID. Leveraging data reduction hardware for this process allows system resources to
be focused on host I/O and other system operations.

Deduplication PowerMax systems use the SHA-256 hashing algorithm implemented in the data
algorithm reduction hardware to find duplicate data. The data is then stored as a single instance for
multiple sources to share. This provides enhanced data efficiency while maintaining a
long history of data integrity.

The SHA-256 algorithm generates a 32-byte code for each 32KB block of data. Consider
a system with 1 PB of written data with 5% updated per day. In 1 million years of
operation, there is a 20% likelihood of a hash collision. As each 128KB track is handled as
4 blocks of 32KB there would need to be a hash collision on all four blocks in the same

8 Dell EMC PowerMax: Data Reduction

Inline compression and deduplication
 Deduplication

128KB track to have an actual hash collision. The odds of having all 4 collide makes this
only theoretical (less than a 1% chance in a trillion years of operation).

Hash table During the data reduction process a hash ID is generated when data is passed through
the data reduction hardware. The hash table stores the unique hash IDs that are used for
comparison as part of the dedupe process. Hash IDs stored in the table are a unique
representation of data in a dedupe relationship. Hash IDs generated by the data reduction
hardware and the SHA-256 algorithm for new writes are compared against the IDs
already populating the hash table. If a matching hash ID already exists in the hash table,
then a dedupe relationship is generated for the newly written data. During the comparison
if the hash ID does not exist the table is updated, and that hash ID is added.

Dedupe The dedupe management object (DMO) is a 64-byte object within system memory. DMOs
Management only exist when dedupe relationships exist. These objects store and manage the pointers
Object between front-end devices and the deduplicated data that consumes backend capacity in
the array.

Data Reduction All I/O is passed through cache and then processed by the system. This means data
I/O flow reduction actions are performed after the data is received by the system, but before it is
placed on disk. Using an inline process requires additional checks within the I/O flow
where data reduction applies. The system uses these checks to determine whether
incoming data needs to pass through the data reduction hardware or not. Incoming data
for a storage group with data reduction enabled will follow the data reduction flow.
However due to the activity-based compression (ABC) function, active data for a storage
group with data reduction enabled will skip the data reduction flow for performance
optimization. Data not compressed due to ABC may be compressed later and moved to a
compression pool. Data for a storage group with data reduction disabled will ignore the
data reduction flow and will be written to the system unreduced.

There are a few different I/O types to consider, Read, Write, Write-update.
• Read - A request to access data that is already populating the array.
• Write - Incoming I/O that will consume disk space.
• Write-update - Incoming I/O that can change data that is allocated to disk space on
the array.

The following figure describes the path the I/O will follow which is determined by
characteristics of the dataset or the related storage group.

Dell EMC PowerMax: Data Reduction 9

Inline compression and deduplication
 Capacity usage

Data reduction
Start Write initiated No
enabled?

Yes

Compress data Is it Active Perform normal

No Yes
and create hash ID Data? I/O flow

Is hash ID in
No
hash table?

Yes

Does a DMO Update hash ID in Add hash ID to

Yes
exist? hash table hash table

No

Are there
<5 Front End Allocate data to
No Create new DMO
Pointers? disk

Yes

Add to existing
Finish
DMO

Figure 1. Data Reduction I/O flow for PowerMax enterprise storage systems

Capacity usage
Data Reduction is a feature intended to offer long-term space savings. Data Reduction
uses machine learning resulting in efficient use of the available system resources. The
use of statistics collected from the incoming data determines what is active and what is
idle. Therefore, the activity-based compression function does not apply to net new writes.
They are compressed and allocated to a compression pool. This also applies to dedupe
as net new writes may not be consuming drive capacity yet. This would be the first entry
of a hash ID into the hash table. Continued access to data generates statistics that are
used to differentiate the activity level of data.

Capacity usage is represented in two ways: a physical capacity used percent and an
effective capacity used percent. While both percentage values reflect how much host
written data is consuming the system, the effected system resource is different. When the
physical capacity usage percent is greater than effective used percent it indicates that
there is potential for usable capacity to reach 100% full. This is also an indication the
achieved data reduction ratio is lower than the systems data reduction reservation. When
the effective used percent is greater than the physical used percent it is an indication
there may be impact to the cache that supports the compression pools. The common
variable related to either percent used is the current data reduction ratio.

10 Dell EMC PowerMax: Data Reduction

Inline compression and deduplication
 Data Reduction Settings

For example, assume a data reduction reservation of 3:1. When the achieved data
reduction ratio is less than 3:1 the physical used percent will be greater than the effective
used percent. Likewise, when the data reduction ratio is greater than 3:1 the effective
used percent will be greater than the physical used percent. Using the same example of
3:1 as the system reservation, applied to 100TB of usable disk capacity. This indicates
the system will manage the resources to achieve 300TB of host data consuming 100TB of
usable capacity. When the achieved ratio is less than 3:1 the system is less likely to
accommodate 300TB of host data on 100TB of usable capacity. When the achieved ratio
is greater than the system reservation the system will accommodate 300TB of host data
on 100TB of disk capacity.

Data Reduction Settings

System and Data reduction is set at the system level per storage resource pool (SRP). The data
Storage reduction reservation set in the system is used to determine the potential savings from
Resource Pool data reduction that can be supported by the available system resources. The data
reduction reservation is used by the system to determine how much cache is needed to
support the potential effective capacity. The cache used to support effective capacity is
allocated as backend metadata in order to support the layout of the storage resource pool.
This also determines how the capacity will be used to store data reduced by data
reduction. As reduced data fills the compression pools, they expand automatically in
order to accommodate more reduced data. Reaching the potential effective capacity relies
on data being written to the system is reducible to a level equal to or greater than the data
reduction reservation. For example, a system with 100TB of usable capacity where the
reservation set is 3.5:1 has a potential effective capacity of 350TB. The written data
needs to be reducible to 3.5:1 or better for the system to accommodate 350TB of host
data and fit it into 100TB of usable capacity.

Storage Group For application workloads to achieve capacity savings from data reduction it must be
enabled at the storage group level. This is supported with Unisphere and Solutions
Enabler. The feature is enabled by default when creating storage groups. There are two
I/O flows for incoming data, data reduction enabled, where data is sent through the data
reduction hardware and reduced, or data reduction disabled in which data bypasses the
data reduction hardware and is written to disk unreduced. The storage group data
reduction setting determines which I/O path the data will follow. In both cases, setting
enabled or disabled is done using the data reduction option when provisioning storage.
The option to enable or disable data reduction for individual storage groups can be
changed at any time. However, changing the setting simply informs the system which I/O
path the data will follow. Changing the setting does not immediately inflate already
reduced data or attempt to reduce data already consuming capacity.

System Efficiency

Overview Data reduction savings are presented as ratios and are available in both Unisphere for
PowerMax and Solutions Enabler. The capacity report provides a single location to view
system efficiency, capacity and system resource usage. The data is displayed in three
sections, Array Usage, Efficiency and System Usage. There are two levels of detail

Dell EMC PowerMax: Data Reduction 11

Inline compression and deduplication
 System Efficiency

available. The default view (see Figure 2) offers a high-level view of efficiency in the form
of ratios and capacity usage displayed as bar graphs. The detailed view expands the
information provided under array usage revealing a more detail of the capacity usage (see
Figure 3). The detailed view also reveals system usage in the form of percentage used
categorized as metadata.

As part of the Q1 2021 PowerMaxOS release there is a further breakdown under the
efficiency section for the data reduction ratio. A flyover display reveals additional
information on the data reduction ratio. The data presented relates specifically to data
reduction enabled allocations. This is divided into two sections, unreducible capacity and
reducible capacity. This information is available in both the default and detailed views of
the capacity report.

• Unreducible Capacity: Represents data reduction enabled allocations the system

was not able to reduce. In most cases this is due to data already compressed
and/or encrypted at the host or application level. In addition, there are data sets
that simply do not reduce well, for example, image files (jpeg, pdf, etc.) or audio
files.
• Reducible capacity: This represents data reduction enabled allocations the
system has determined are reducible. The values are displayed as capacity
representing reducible data that will be reduced. Adding compression and dedupe
savings with pattern detection savings may not equal the reducible capacity value
due to performance optimization (see Activity Based Compression). The enabled
for data reduction percentage will also be a factor when less than 100.

Figure 2. System efficiency report as seen in Unisphere for PowerMax (Default high-level
view).

Calculating Efficiency Ratios: This data is revealed in the array usage section when
switching the capacity report to the detailed view (see Figure 3). It is needed to calculate
the ratios displayed in the efficiency section. The data available from the detailed view
can be used in the formulas below to calculate the ratios displayed in the efficiency
section.

12 Dell EMC PowerMax: Data Reduction

Inline compression and deduplication
 System Efficiency

Figure 3. System efficiency report detailed view of Array Usage in Unisphere for
PowerMax

• Overall Efficiency Ratio: The range of values that describe the capacity space
savings that a user may experience regarding data reduction or other data services
that offer capacity savings, such as Data Reduction, non-zero allocation, over
provisioning and SnapVX.

𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 + 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆ℎ𝑜𝑜𝑜𝑜 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡

𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈

• Data Reduction Ratio: Savings that represents the combination of inline

compression and inline deduplication presented as a ratio. To calculate the data
reduction ratio the user needs to toggle to the detailed view shown in Figure 3.
Additional information needed is revealed by hovering over the Usage bars. When
calculating the data reduction ratio using the values presented in the usage portion
of the capacity report. The ratio may reflect a different value due to the performance
optimization leaving compressible data uncompressed. The enabled percent being
less than 100 may also be a factor.
𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝒃𝒃𝒆𝒆𝒆𝒆 𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨 𝑵𝑵𝑵𝑵𝑵𝑵 𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺 + 𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴 𝑵𝑵𝑵𝑵𝑵𝑵 𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺
𝑼𝑼𝑼𝑼𝑼𝑼𝑼𝑼 𝑼𝑼𝑼𝑼𝑼𝑼𝑼𝑼

• Data Reduction Ratio on Reducible: Represents the data reduction savings using
only data reduction enabled allocations that have been reduced.
𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍𝒍
𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹 − (𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪 + 𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷)

• Enabled Percent: The amount of subscribed host allocations that have Data
Reduction enabled.
• Virtual Provisioning Savings: Savings achieved relative to provisioned capacity
and total usable capacity displayed as a ratio. This may exceed the maximum
usable capacity.
𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺 𝑻𝑻𝑻𝑻𝑻𝑻𝑻𝑻𝑻𝑻 𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪
𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨 𝒏𝒏𝒏𝒏𝒏𝒏 − 𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔

Dell EMC PowerMax: Data Reduction 13

Inline compression and deduplication
 System Efficiency

• Snapshot Savings: A representation of savings resulting from the use of SnapVX

to create local replication data.
𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺 𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪 𝑻𝑻𝑻𝑻𝑻𝑻𝑻𝑻𝑻𝑻
𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴 𝑵𝑵𝑵𝑵𝑵𝑵𝑵𝑵𝑵𝑵𝑵𝑵𝑵𝑵𝑵𝑵𝑵𝑵

System In this section system resource usage refers to the two main components of the system
Resource Usage relative to data reduction, capacity and cache.

Capacity is displayed as subscribed, and usable. Subscribed capacity represents the

amount of capacity presented to hosts or applications. Subscribed capacity exceeding
the total usable capacity is commonly referred to as subscription or virtual provisioning.
This allows users to present more capacity to applications or hosts than the system can
store. This is displayed in Unisphere as the Subscribed Usable Capacity Percent.
Usable capacity is the amount of disk capacity available to store application data. The
total amount of usable capacity is determined by the capacity of the physical disks that
are configured in the system.

Like previous generations, PowerMax operates as a cache centric architecture. All data is
passed through cache before being stored on disk. It is used to support multiple functions
within the system not simply host I/O. Provisioning, local replication and data reduction
also use cache. Cache is divided into two main sections, data cache and metadata
cache. (see Figure 4). Cache usage is displayed in the Unisphere capacity report under
the System usage section, represented as metadata used (see Figure 6).

• Data cache: Represents the amount of cache available for host IO, reads and
writes from hosts or applications. The system configuration ensures there is always
data cache available for host I/O.
• Metadata cache: This is comprised of three sections, Front-End, Replication and
Back-End. Each section represents an amount of metadata cache it can consume.
 Front-End Metadata: At initial install the percentage used will show zero as
there is no subscribed capacity. There are two factors that will cause front-end
metadata usage to increase, provisioning capacity to hosts or applications as
well as host allocations. In PowerMax systems the increase is primarily due to
host allocations.
 Replication Metadata: At initial install the percentage used will show zero as
there is no local replication activity. As local replication is used the percentage
will increase up to 100 percent. When replication metadata has reached 100
percent, local replication has reached its limit. (for more information refer to
dell-emc-powermax-vmax-all-flash-timefinder-snapvx-local-replication.pdf)
 Back-End Metadata: At initial install the percentage used represent the initial
layout of compression pools. As the compression pools are expanded to
support more effective capacity the usage can grow up to 100%. When Back-
End Metadata shows 100% used it indicates the system has expanded usable
capacity to the maximum effective capacity the system can support. This has
no impact on Front end metadata growth or the ability to support Host I/O.

14 Dell EMC PowerMax: Data Reduction

Inline compression and deduplication
 System Efficiency

Figure 4. Cache breakdown

Array usage:

• Subscribed Capacity: The bar presented represents the total provisioned

capacity. The dark shaded portion of the bar indicates host allocations of the
presented capacity.
• Snapshot Capacity: The total represents the sum of all existing snapshots. The
dark shaded portion of the bar represents the amount of existing snapshot capacity
that has been modified. The modified capacity also represents additional usable
capacity that is consumed by snapshot data.
• Usable Capacity: The total amount of usable disk space available. The dark
shaded portion represents the amount of disk space that is consumed.
• Subscribed Usable Capacity: The percent displayed represents the amount of
subscribed capacity in relation to the total amount of usable capacity.

Figure 5. Array usage from the Unisphere for PowerMax Capacity report

Dell EMC PowerMax: Data Reduction 15

Inline compression and deduplication
 System Efficiency

System Usage: The capacity report in Unisphere for PowerMax displays meta data
usage in the form of percentage used. The values displayed represent the amount of
metadata used for each function. These values are also available in Solution Enabler and
REST API.

Figure 6. System Usage as seen in Unisphere for PowerMax

• System Metadata: Represents the total metadata usage for the system. The
amount of cache used by the system for all functions supported by metadata. The
system used percentage represents the usage encompassing the total amount of
metadata cache available.
 Replication Metadata: A cache resource used in the form of metadata to
support replication data pointers used with local replication. At initial install the
percent used starts at zero as there is no local replication activity. The total
amount of cache available for replication metadata is based on the
configuration of the system and will not increase with the use of local
replication. When replication metadata has reached its maximum the use of
local replication has reached its limit.
 Front-End Metadata: A cache resource used in the form of metadata to
support subscribed capacity and host allocations. As subscribed capacity is
increased, the amount of front-end metadata increases. In VMAX All Flash
systems provisioning will cause this to increase at the time of device creation.
In PowerMax system the increase is primarily due to host allocations. In both
cases the increase of Front-End metadata can consume data cache.

Storage group Data reduction savings displayed at the Storage Group level represents only compression
compression in savings. This information can be viewed with the Storage Group list, the detailed view
Unisphere and the Storage Group Demand report. The ratio displayed shows the compression
savings for data specific to the storage group being viewed. In addition to the
compression ratio the amount of unreducible data is shown. The amount of unreducible
data shown represents the amount of data the Storage group has allocated that the
system has determined is not reducible. See the examples in the following figures.

Figure 7. Storage group list view in Unisphere for PowerMax

16 Dell EMC PowerMax: Data Reduction

Inline compression and deduplication
 Supported data services

Figure 8. Storage group details view in Unisphere for PowerMax

Figure 9. Storage group details > volume tab view in Unisphere for PowerMax

Figure 10. Storage group demand report in Unisphere for PowerMax

Supported data services

Introduction Data reduction is supported for FBA storage. Mixed FBA/CKD systems are supported
however data reduction will only apply to the FBA storage resource pool/s. All other data
services offered in both the PowerMax and VMAX All Flash systems are supported. This
includes local replication (SnapVX), remote replication (SRDF), D@RE, and VMware
vSphere Virtual Volumes (vVols).

Local replication Data reduction is supported with the use of local replication features; there are multiple
(SnapVX) variations and use cases for local replication. Below are the details regarding the different
local replication sessions that can exist. For more detail regarding local replication and
SnapVX see the TimeFinder and HYPERMAX OS Local Replication Technical Note
available at DellEMC.com.

Nocopy session (SnapVX, VP Snap)

Uncompressed source data remains uncompressed when becoming snapshot data and
may be compressed later as it becomes less active. Activity to snapshot data through a
linked target may prevent uncompressed data from being compressed. Compressed
source data remains compressed when becoming snapshot data. Read activity to a
snapshot through a linked target may cause the compressed data to be uncompressed.

Dell EMC PowerMax: Data Reduction 17

Inline compression and deduplication
 Conclusion

The compression setting of a linked target only affects data written directly to the linked
target and does not affect the snapshot data.

Copy sessions (SnapVX Full Copy Linked Targets, Clone, Mirror)

The compression settings for both the source and target are taken into account for copy
sessions.

When compression is enabled on the source the data is decompressed before copying to
the target. When compression is enabled on the target the data is compressed before
being allocated to the target. Likewise, when compression is enabled on both the source
and target the data is decompressed before the copy and then compressed to allocate for
the target.

Copy times may vary due to decompression and compression of the data. It is not
recommended to change the compression settings in between differential operations (that
is, disabling compression before each differential operation and then again after the copy
completes) as this causes data to go through needless compress/decompress cycles.

Remote Compression for SRDF is already supported and known as SRDF compression. SRDF
replication compression is a feature designed to reduce bandwidth consumption while sending data
(SRDF) to and from systems connected using remote replication. SRDF compression and the
Adaptive Compression Engine (ACE) both use the same compression module; however,
they serve different purposes. Data that has been compressed using ACE is
uncompressed before being sent across the SRDF link. If SRDF compression and inline
compression apply, the data is uncompressed by the module and then compressed using
the SRDF compression function and then sent to the remote site.

Data at Rest D@RE provides hardware-based, on-array, back-end encryption, Data Reduction
Encryption provides inline compression and Deduplication. Data is passed through the Data
(D@RE) Reduction hardware before being sent through the encryption hardware. Therefore, data
is compressed, deduped, or both before being encrypted by the D@RE process. On a
D@RE enabled system data encrypted on disk has already been compressed, deduped,
or both.

Virtual Volumes Data reduction is supported for the allocation of data to vVols and follows the same I/O
path as all other data. The IO path can be seen in Figure 7. Data Reduction as a feature
is not included as a vVols resource to be configured at the host.

Conclusion
The use of physical storage capacity is a common concern of storage administrators
across the storage industry. The constant and ever-growing amounts of data have created
the need for more efficiency in the use of physical capacity. Dell EMC PowerMax and
VMAX All Flash data storage systems take this to the next level. Combining inline
compression with inline dedupe provides exceptional capacity savings with negligible cost
to performance. This delivers on capacity savings, which leads to a smaller data center
footprint and an overall reduction in TCO. In addition to the savings, using data reduction
is as simple as a single click to enable or disable. The system handles all the work.

18 Dell EMC PowerMax: Data Reduction

Inline compression and deduplication
 Appendix: Technical support and resources

Appendix: Technical support and resources

Dell.com/support is focused on meeting customer needs with proven services and
support.

Storage and data protection technical white papers and videos provide expertise that
helps to ensure customer success with Dell EMC storage and data protection platforms.

Dell EMC PowerMax: Data Reduction 19

Inline compression and deduplication