Table of Contents
- EXECUTIVE SUMMARY 5
- INTENDED AUDIENCE 5
- The VMAX All Flash Family 5
- FLASH OPTIMIZATION ON THE VMAX ALL FLASH 11
- VMAX ALL FLASH DATA SERVICES 12
- VMAX All Flash – High Availability and Resilience 14
- VMAX All Flash Configurations for Open Systems 15
- VMAX All Flash Configurations for Mainframe 19
- SUMMARY 23
- REFERENCES 23
- EXECUTIVE SUMMARY
- INTENDED AUDIENCE
- The VMAX All Flash Family
- FLASH OPTIMIZATION ON THE VMAX ALL FLASH
- VMAX ALL FLASH DATA SERVICES
- VMAX All Flash – High Availability and Resilience
- VMAX All Flash Configurations for Open Systems
- VMAX All Flash for Mainframe
- SUMMARY
- REFERENCES
EMC VMAX 850FX User Manual
Displayed below is the user manual for VMAX 850FX by EMC which is a product in the Mainframe Computers category. This manual has pages.
Related Manuals
WHITE PAPER
THE VMAX AL L F LASH STORAGE FAMILY
A Detailed Overview
ABSTRACT
Recent engineering advancements with higher density, vertical, multi-layer cell flash
technology have led to the development of higher capacity multi-terabyte flash drives.
The introduction of these higher capacity flash drives has greatly accelerated the
inflection point of where flash drives have the equivalent economics of traditional hard
drives functioning as the primary storage media for enterprise applications in the data
center. DELL EMC engineers foresaw this inflection point and are now pleased to
introduce the VMAX® All Flash family. This w hite pap er provides the reader with an
in-depth overview of the VMAX All Flash family by providing details on the theory of
operation, packaging, and the unique features which make it the premier all-flash
storage product for the modern data center.
September 2016
2
The information in this publication is provided “as is.” EMC Corporation 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 EMC software described in this publication requires an applicable software license.
EMC2, EMC, the EMC logo, are registered trademarks or trademarks of EMC Corporation in the United States and ot her coun tries. All
other trademarks used herein are the property of their respective owners. © Copyright 2016 EMC Corporation. All rights reserved.
Published in the USA. 09/16, white paper, Part Number H14920.1
EMC believes the infor mat ion i n this docu ment is accurate as of its publication date. The information is subject to change without
notice.
EMC is now part of the Dell group of companies.
3
TABLE OF CONTENTS
EXECUTIVE SUMMARY ........................................................................................................................................................ 5
INTENDED AUDIENCE .......................................................................................................................................................... 5
THE VMAX ALL FLASH FAMILY .......................................................................................................................................... 5
Background ......................................................................................................................................................................... 5
Introducing the VMAX All Flash Arrays .............................................................................................................................. 5
VMAX All Flash System Overview ...................................................................................................................................... 6
A Modular Building Block Architecture ............................................................................................................................... 6
The Brick Overview ............................................................................................................................................................. 7
Brick Engines ...................................................................................................................................................................... 8
Brick Engine CPU Core Configurations .......................................................................................................................... 8
Brick Engine Cache Configurations ................................................................................................................................ 8
Brick Drive Array Enclosures (DAEs) ................................................................................................................................. 9
VMAX 250F Model V-Brick DAEs................................................................................................................................... 9
VMAX 450F and VMAX 850F Model Brick DAEs........................................................................................................... 9
Important notes about VMAX All Flash brick DAE capacity ......................................................................................... 11
FLASH OPT IMI Z ATION ON THE VMAX ALL FLASH ........................................................................................................ 11
VMAX All Flash Cache Architecture and Caching Algorithms .......................................................................................... 11
Understanding Flash Cell Endurance ............................................................................................................................... 11
VMAX All Flash Write Amplification Reduction................................................................................................................. 12
Boosting Flash Performance with FlashBoost .................................................................................................................. 12
HYPERMAX OS ............................................................................................................................................................... 12
VMAX ALL FLASH DATA SERVICES ................................................................................................................................ 12
Remote Replication with SRDF ........................................................................................................................................ 13
Local Replicat io n with T imeFinder SnapVX ..................................................................................................................... 13
Consolid ati on of Block and File Storage U sin g eNAS ...................................................................................................... 13
Embedded Management (Embedded Unisphere for VMAX) ........................................................................................... 14
VMAX ALL FLASH – HI GH AVAILABIL ITY AN D RESILIENCE ........................................................................................ 14
VMAX ALL FLASH CONFIGURATIONS FOR OPEN SYSTEMS ....................................................................................... 15
Open Systems V-Brick System Configurations for the VMAX 250F ................................................................................ 15
Open Systems V-Brick System Configurations for the VMAX 450F and VMAX 850F .................................................... 16
Open Systems V-Brick Front End Connectivity Options .................................................................................................. 16
Open System VMAX All Flash Software Packaging ......................................................................................................... 18
VMAX ALL FLASH CONFIGURATIONS FOR MAINFRAME ............................................................................................. 19
Mainframe zBrick System Configurations for the VMAX 450F / 850F ............................................................................. 19
Mainframe zBrick Front End Connectivity Options ........................................................................................................... 20
VMAX All Flash for Mainframe Software Packaging ........................................................................................................ 22
4
SUMMARY ............................................................................................................................................................................ 23
REFERENCES ...................................................................................................................................................................... 23
5
EXECUTIVE SUMMARY
In 2016, an inflection point has been reached where flash storage now has the equivalent density and econom ic s of traditi onal spi nni ng
Hard Disk Drive (HDD) media. This inflection point has fundamentally changed the data storage landscape in the enterprise data
center. To meet the demands of the all flash enterprise storage environment, DELL EMC is pleased to offer an all-flash offering called
the VMAX All Flash family.
The VMAX All Flash family expands the conversation in the all-flash space to include mission-critical resiliency, native and trusted
enterprise data services, and consolidation of workloads beyond block storage. VMAX All Flash sets itself apart from the other
competitors in the enterprise flash storage space as it provides customers with:
• A trusted architecture delivering unmatched 6x9’s availability for enterprise-level flash storage requirements
• The most trusted data services in the industry, including SRDF and Timefinder SnapVX - the gold standards for remote and
local replication technologies
• Unmatched flash density per floor tile with both block and file workloads co-existing within the same system
VMAX All Flash offers unmatched simplicity for customers when it comes to planning, ordering, and management. There are three
VMAX All Flash models: the VMAX 250F, the VMAX 450F, and the VMAX 850F. Customers can scale up and scale out using a simple
modular architecture. Each VMAX All Flash model can be ordered with a pre-packaged software bundle – the entry “F” package, or the
more encompas sing “FX” package. Every VMAX All Flash model comes standard with embedded Unisphere for VMAX for easy and
simple storage man age men t and monit oring. VMAX All Flash also delivers unmatched simplicity for maintenance and licensing which
helps dramatically lowers a customer’s Total Cost of Ownership (TCO) of the product.
The VMAX All Flash family has truly changed the lands cape of the ent erpr i se data cen ter. The primary compone nts and benefits of this
game changing product will be discussed in detail in the following sections of this document.
INTENDED AUDIENCE
This white paper is intended for DELL EMC customers and p otential custo mers , DELL EMC Sales and Support Staff, Partners, and
anyone inte rested in gaining a better understanding of the VMAX All Flash storage array and its features.
The VMAX All Flash Family
Background
Enterprise storage capacity and storage performance requirements have increased dramatically in recent years with the need to
support million s of virtual devices and virtual machines. Although traditional spinning disk media can still meet the storage capacity
requirements, it is having difficulty meeting the performance requirements (now measured in the millions of IOPS) for these
environments.
Until recently, the industry was in a quandary as the economics of all flash storage were still prohibitive; however, the recent
advancements in flash technology – specifically the development vertical, 3-bit, charge trap NAND architectures - have led to a
breakthrough in the capacity and economics of flash storage. This breakthrough has greatly accelerated the inflection point of where
flash storage has the same economics of traditional spinning disk media. The release of these new drives is now allowing the
enterprise data center to meet the storage capacity and performance requirements for highly virtualized environments at affordable
economics.
Introducing the VMAX All Flash Arrays
To meet the emerging requirements of the enterprise storage environment, DELL EMC is pleased to introduce a new all flash offering
called VMAX All Flash. The VMAX All Flash will offer three distinct base models- the VMAX 250F, the VMAX 450F, and the VMAX
850F The VMAX All Flash arrays have at their foundation the trusted Dynamic Virtual Matrix architecture and HYPERMAX OS;
however, they differ from the VMAX hybrid arrays as they are true all flash arrays – being the VMAX products specifically targeted to
meet the storage capacity and performance requir em ents of the all flash enterprise data center. The VMAX All Flash products are
feature-rich all flash offerings with specific capabilities designed to take advantage of the new higher capacity flash drives used in the
densest configuration possible. The VMAX All Flash arrays offer enterprise customers the trusted VMAX data services, the improved
simplicity, capa city, and performance which thei r highly virt u aliz ed env iron ment s dema nd, while still meeting the eco nom i cs of the more
traditional storage workloads.
6
The VMAX All Flash product line has been engineered to deliver on the following key design objectives:
• Performance – Regardless of workload and regardless of storage capacity utilization, VMAX All Flash is designed to provide
consistently predictable high performance to the enterprise data center, delivering up to 4 million IOPS with less than 0.5 ms
latency at 150 GB/sec bandwidth.
• High Availability and Resiliency – VMAX All Flash is built with trusted architecture which bears no single points of failure
and has a proven 6x9’s availability track record. The ability to use of SRDF gives customers full multi-site repli cat ion opt i o ns
for disaster recovery and rapid restart.
• Inline Compression - Compression is a space saving function designed to allow the HYPERMAX OS to manage capacity in
the most efficient way possible. Compression is performed by the HYPERMAX OS within the system using mul tipl e
compression ranges in order to achieve the 2:1 average for the system.
• Non-Disruptive Migration (NDM) - NDM is designed to help automate the process of migrating hosts and applications to a
new VMAX All Flash array with no downtime at all.
• Enhancing Flash Drive Endurance –VMAX All Flash has unique capabilities to greatly minimize write amplification on the
flash drives. It employs large amounts of cache to store writes and then uses intelligent de-stage algorithms to coalesce the
writes into a lar ger seq uent ial w rite – minimizing random write I/O to the back end. VMAX All Flash also employs proven write
folding algorithms which drastically reduce the amount of write I/O to the back end.
• Flash Density – Using high capacity flash drives, VMAX All Flash will deliver the highest IOPS/TB/floor tile in the industr y .
VMAX All Flash support for high capacity flash drives provides a differentiated capability versus many all flash alternatives. It
allows the system to leverage the increases in flash drive densities, economies of scale, and fast time to market provided by
the suppliers for industry standard flash drive technology.
• Scalability – VMAX All Flash configurations are built with modular build ing blo ck s called “Bricks”. A Brick includes an engine
and two drive DAEs pre-configured with an initial total usable capacity. Brick capacity can be scaled up in specific incr em e nts
of usable capacity called Flash Capacity Packs.
• Data Services – Full support for the industry’s gold standard s in remote replication with SRDF and local replication with
Timefinder SnapVX. VMAX All Flash will also have a full integration with DELL EMC AppSync for easier local replication
management of critical applications.
• Consolidation – The VMAX All Flash are the only all flash storage products in the industry which can consolidate open
system block and f ile storage onto a single floor tile. VMAX All Flash supports many front end conn ect iv ity options including
Fibre Channel, iSCSI, and FICON for Mainframe
• Streamlined Packaging - The VMAX All Flash family450F850Fwill feature an “F” and “FX” options.. The difference in the
models is specifically related to the greatly simplified software packaging for the VMAX All Flash product line. The VMAX All
Flash base models will always be referred to as the VMAX 250F, 450F, and 850F. The base “F” model will offer an entry level
software packaging which will include features such as embedded Unisphere; while the “FX” includes the entry level “F”
packaging, plus more advanced software offerings such as SRDF.
• Ease of Managem ent – Embedded Unisphere for VMAX is provided in both the F and FX package. The Unisphere for VMAX
intuitive management interface allows IT managers to maximize productivity by dramatically reducing the time required to
provision, manage, and monitor VMAX All Flash storage assets. The fact that Unisphere for VMAX is embedded within VMAX
All Flash allows for this simplicity of management without the need for additional servers and hardware. The FX package also
includes Unisphere 360, which provides storage administrators the ability to view site-level health reports for every VMAX in
the data center, and also coordinate compliance to code levels and other infrastructure maintenance requirement
VMAX All Flash System Overview
VMAX All Flash is architected to support the densest flash configuration possible. VMAX All Flash support for high capacity flash drives
provides a differentiated capability versus many all flash alternatives. It allows VMAX All Flash to leverage the increases in flash drive
densities, eco nom ie s of scale, and faster time to market provided by the suppliers for industry standard flash drive technology.
The shift to higher capacity flash drives provides an attractive alternative to hybrid arrays in terms of acquisition cost and overall total
cost of ownership. Other VMAX All Flash advantages such as higher performance, predictable latency, increased density, reduced
power and cooling, and reduced drive replacement will accelerate deployment of all flash based VMAX storage systems going forward.
A Modul ar Building Block Architecture
VMAX All Flash employs a simplified appliance based software packaging and modular building block configuration to reduce
complexity to make it easier to configure and deploy. This architecture allows it to scale to deliver predicable high performance where
needed. These building blocks are called “bricks”.
7
There are two types of bricks available for the VMAX All Flash:
• The V-Brick which supports open system configurations with Fibre Chan nel and / or iSCS I connect iv ity and FBA devi ce
formatting
• The zBrick which supports Mainframe configurations with FICON connectivity and CKD device formatting.
Note: In this document, the term “brick” will be used when discussing features and functions applicable to both the V-Brick and the
zBrick. The zBrick will be discussed in more detail in the VMAX All Flash Mainframe support portion of this document.
The Brick Overview
The core element of VMAX All Flash is the brick. Each brick has the following components:
• One engine using the dynamic virtual matrix architecture running HYPERMAX OS
• Fully redundant hardware with multiple power supplies and interconnecting fabrics
o No single points of failure architecture
o Proven 6x9s availability
• 2 x 2.5” drive slot Drive Array Enclosures (DAEs)
o VMAX 250F has 2 x 25 slot 2.5” drive, VMAX 450F / 850F has 2 x 120 slot 2.5” drive
o VMAX All Flash starter brick configuration has a set amount of usable capacity
o Additional V-Brick storage capacity is added in defined increments called “Flash Capacity Packs”, while add itio nal
zBrick storage capacity is added in defined increments called “zCapacity Packs”.
• Up to 32 ports of front end connectivity
• Up to 2 TB of Cache per Brick
The following table details the various VMAX All Flash model brick specifications:
Table 1. Brick Specifications by VMAX All Flash Model
Component
Specification
VMAX 250F
VMAX 450F
VMAX 850F
System Layout
Floor Tile Space Required
1
1 - 2
1 - 4
Compute
# of Bricks per System
1 - 2
1 - 4
1 - 8
Support for Mainframe zBrick
No
Yes
Yes
Maximum # of cores per system
96
128
384
Cache
Cache per Brick Options
512 GB, 1 TB, and 2 TB
1 TB and 2 TB
Mixed cache support
Yes
No
No
Ports and
Modules
Maximum FE modules per V-Brick
8 (32 total FE ports per V-
Brick)
6 (24 total FE ports per V-Brick)
Maximum FE modules per zBrick
NA
6 (24 total FICON ports per zBrick) (1)
Maximum FE ports per system
64
96
192
Drives and
Capacity
Brick DAE Type and QTY
2 x 25 slot, 2.5" (DAE25)
2 x 120 slot, 2.5"
Maximum # of drives per system
100
960
1920
Maximum open systems effective
capacity per system (1)
1 PBe
2 PBe
4 PBe
Maximum mainframe usable capacity per
system (2)
NA
800 TBu
1.7PBu
Starter Brick usable capacity
11 TBu
53 TBu
Flash Capacity Pack increment size
11 TBu
13 TBu
RAID Options
RAID 5 (3+1),
RAID 6 (6+2)
RAID 5 (7+1), RAID 6 (14+2)
Supported V-Brick Flash Drive Sizes
960 GB, 1.92 TB, 3.84 TB,
7.68 TB, 15.36 TB
960 GB, 1.92 TB, 3.84 TB
Supported zBrick Flash Drives Sizes
NA
960 GB, 1.92 TB, 3.84 TB
8
(1) Default zBrick comes with 2 FICON modules. Extra FICON modules can be ordered al la carte
(2) DELL EMC uses PBu (and TBu) to define usable storage capacity in the absence of compression, i.e. referring to the
amount of physical storage in the box. DELL EMC uses PBe (and TBe) to define effective storage capacity in the
presence of compre ssi on,
(a) i.e. if a customer has 50TBu of physical storage, and it is compressible on a 2:1 basis, then the customer has
100TBe (Effe ctive stora ge).
The brick concept allows VMAX All Flash to “scale up” and “scale out”. Customers can scale up by adding Flash Capacity Packs. Each
Flash Capacity Pack has a multipl e of 13 TBu of usable stora ge for the VMAX 450F / VMAX 850F models, and 11 TBu for the VMAX
250F model. VMAX All Flash scales out by aggregating up to two bricks for the VMAX 250F, and up to eight bricks for the VMAX 450F /
850F in a single system with fully shared connectivity, processing, and capacity resources. Scal ing out a VMAX All Flash system by
adding additional bricks produces a predictable, linear performance improvement regardless of the workload.
Brick Engines
The brick engine is the central I/O processing unit, redundantly built for high availability. It consists of redundant directors that each
contain multi-core CPUs, memory modules, and attach interfaces to universal I/O modules, such as front-end, back-end, InfiniBand,
and flash I/O modules.
The foundation of the brick engine is the trusted Dynamic Virtual Matrix Architecture. Fundamentally, the virtual matrix enables inter-
director commun ic ations over redund ant inter nal Inf i niBa nd fabr ic s. The Infin iBand fabric pr ov ides a founda tion for a highly scalable,
extremely low latency, and high bandwidth backbone which is essential for an all flash array. This capability is also essential for
allowing the brick concept to scale upwards and scale outwards in the manner that it does.
Brick Engine CPU Core Configurations
Each brick engine has two directors, with each director having dual CPU sockets which can support multi-core, mul ti-threaded Intel
processors. The following table details the engine CPU core layout for each VMAX All Flash model:
Table 2. Brick Engine CPU Cores per VMAX All Flash Model
VMAX All Flash
Model
Engine CPU Type
Cores per Director
Cores per Brick
Max Cores per System
250F (V-Brick Only)
Dual Intel Broadwell 12 core
24
48
96 (2 bricks max.)
450F
Dual Intel Ivy Bridge 8 core
16
32
128 (4 bricks max.)
850F
Dual Intel Ivy Bridge 12 core
24
48
384 (8 bricks max.)
The brick engine uses a core pooling mechanism which can dynamically load balance the cores by distributing them to the front end,
back end, and data services (such as SRDF, eNAS, and embedded management) running on the engine. The core pools can be
dynamicall y tuned to shift the bias of the pools at any time to front end heavy or back end heavy workloads to further optim i ze the
solution for a specific use case.
Aside from being able to dynamically adjus t the core pools , VMAX All Flash has the capability to implement advanced Quality of
Services (QoS), such as setting the maximum amount of IOPS for a particular storage group. This is extremely helpful at properly
managing system core consumption so that a “noisy” virtual machine or host cannot overly consume system resources. QoS helps
ensure that all connected hosts and virtual machines receive an evenly distributed amount of res ources to deliver the ma ximum
performanc e possible in terms of IOPS and throughput.
Brick Engine Cache Configurations
Every brick director has 16 memory slots whi ch can be populated with 32 GB and 64 GB DDR4 DIMMS to achieve up to 1 TB cache
per director (2 TB cache maximum per brick engine).
Table 3. Brick Engine Cache Configuration per VM AX All Flash Model
VMAX All Flash Model
Cache per Brick
Max Cache per S ystem
250F (V- Brick Only)
512 GB, 1 TB, 2 TB
4 TB (2 bricks max.)
450F
1 TB or 2 TB
8 TB (4 bricks max.)
850F
1 TB or 2 TB
16 TB (8 bricks max.)
9
For dual brick VMAX All Flash 250F models, the system can us e engines with differing cache sizes (m i xed memory). For example, the
cache for the engine on br ick A can be 1 TB while the cache for the engine for brick B can be 512 TB. This would yield a total system
cache size of 1.5 TB. The difference in cache sizes between the engines has to be one capacity size smaller or larger. Valid mixed
memory configur at i ons for the VMAX 250F are shown in the following table:
Table 4. VMAX 250F Mixed Engine Cache Size Configurations
Mixed Memory Configuration
Smallest Engine Cache Size
Largest Engine Cache Size
Total System Cache
Configuration 1
512 GB
1 TB
1.5 TB
Configuration 2
1 TB
2 TB
3 TB
Note: The VMAX 450F and 850F models do not support m i xed cache sizes between engines. In these systems, the cache size
between engine s must be equal.
The VMAX All Flash family supports Dynamic Cache Partitioning (DCP) on the system engines. DCP is a QoS feature which allows for
the fencing off of specific amounts of cache for particular environments such as “production” from “development”. Another example
would be the separation of c ache resources for “file data” from “block data” on systems which are using eNAS services . Being able to
fence off and isolate cache resources is k ey enabler for multi-tenant environments.
Brick Drive Array Enclosures (DAEs)
VMAX 250F Model V-Bri ck DAEs
Each brick for the VMAX 250F c omes with two 25 slot, 2.5” drive, 2U front loading DAEs along wi th 11 TBu of pre-c onfigured initial
capacity that can use either RAID 5 3+1 or RAID 6 6+2 protection. The VMAX 250F DAE supports 12 Gb/sec SAS connectivity and
requires 12 Gb/sec SAS flash drives. Flash drives which use 6 Gb/sec SAS c onnectivity are not supported in the VMAX 250F. The
VMAX 250F DAE has dual ported drive slots and dual power zones for high availability.
Figure 1. Fully populated VMAX 250F V-Brick DAE
Additional scale up capacity is added to the VMAX 250F system using flash capacity pack increments of 11 TBu, scal ing up to a
maximum of effective capacity of 500 TBe per brick. A dual brick VMAX 250F can sc al e up to a total capacity 1PBe using a half rack
(20U) within a single floor tile footprint.
VMAX 450F and VMAX 850F Model Brick DAEs
Each brick for the VMAX 450F and VMAX 850F comes with two 120 slot, 2.5” drive, 3U drawer DAEs, along with 53 TBu of pre-
configure d initi al capa city t hat can use eit her RAID 5 7+1 or RA ID 6 14+2 protect ion. Each VMAX 450F and VMAX 850F brick DAE
uses dual ported drive slots and uses four separate power zones to eliminate any single points of failure. The VMAX 450F and VMAX
850F brick DAE layout is shown in the diagram below:
10
Figure 2. VMAX 450F / 850F Brick DAE Layout – Drive Slots and Power Zones
Additional scale up capacity is added to the system in multiples of 13 TBu increments called “Flash Capacity Pack s ” for V-Bricks and
“zCapacity Packs” for zBricks. The capacity pack concept allows for considerable internal capacity growth over the lifespan of the
VMAX 450F and VMAX 850F arrays – especially when higher capacity flash drives are used. Each VMAX 450F and VMAX 850F
system can start small with as little as 53 TBu of capacity then scale to up to 500 TBu of usable capacity using a single 2 TB cache
engine. The additional capacity can be added into the empty slots of the DAEs. This allows for easy expansion as no extra DAEs would
need to be added into the sy st em bay . When a two brick VMAX 450F or VMAX 850F system bay is deployed with 2 TB cache engi nes
using (2:1) compression, a c ustomer can have up to 1 PBe of effective flash capacity on a single floor tile whi le using only 500 TB of
physical stora ge. U sin g an ave r age of 2:1 compression, the VMAX 450F system can scale to 2 PBe and the VMAX 850F system can
scale to 4 PBe.
Figure 3. Fully populated VMAX 450F / 850F Brick DAE
11
Important notes about VM AX All Flash brick DAE capacity
• VMAX All Flash arrays will use a single RAID protection scheme for the entire system. The spe cifi c protection scheme is
determined by the initial usable capacity of the system. All follow on capacity and brick additions will use the sam e RAID
protection scheme as the initial usable capacity (53 TBu for the VMAX 450F and VMAX 850F, 11 TBu for the VMAX 250F)
regardless of the drive size used by the additional flas h capacity pack.
• VMAX All Flash addressable capacity, space available for host IO, is governed by the amount of total cache in the system.
Typically, 1 TB of V-Brick engi ne cache can usually support up to 250 TB of open system host addressable storage; while 1
TB of zBrick engine cache can usually support up to 100 TB of mainframe host addressable storage. This becomes important
in properly sizing a VMAX All Flash system with compression. As an example, if a customer requires having 1 PBe of
addressable storage compressed at a 2:1 ratio, this means that the system will require 4 TB of system cache and 500 TB of
physical stora ge.
• M ultiple flash driv e sizes can co-exist within the brick DAE for the VMAX 250F, VMAX 450F, and VMAX 850F.
• The brick RAID groups span across both DAEs
• The VMAX All Flash will offer the “Diamond” Service Level for internal storage and the “Optimized” Service Level for external
disk storage such as Cloud Array.
• Spare drive requirements are calculated with 1 spare per 50 drives of a particular type, on a per engine basis
• Mai nframe zBrick DAEs will support 960 GB, 1.92 TB and 3.84 TB flash drives
FLASH OPTIM IZATION ON THE VMAX ALL FLASH
All flash based storage systems demand the highest levels of performance and resiliency from the enterpris e storage platforms that
support them. The foundation of a true all flash array is an architecture which can fully leverage the aggregated performance of modern
high density flash drives while maximizing their useful life. VMAX All Flash has several features built into the architecture of the product
specifical ly designed to maximize flash dri ve performance and longevity. This section will discuss these features in detail.
VMAX All Flash Cac he Architecture and Caching Algorithms
The VMAX family is built upon a very large, high-speed DRAM cache based architecture, driven by highly complex and optimized
algorithms. These algorithms accelerate data access by avoiding physic al access to the back end whenever poss ible.
DELL EMC has spent many years developing and optimizing caching algorithms. The VMAX cache algorithms optimize reads and
writes to maximize I/Os serviced from cache and minimize access to back end flash drives. The system also goes to lengths at trying to
predict what data may be needed in the future by the applications by an IO’s l ocality of reference. This data is also pulled up into cache.
Some of the techniques used by the cache algorithms to minimize disk access are:
• 100% of host writes are cached
• More than 50% of reads are cached
• Recent data is held in cache for long periods, as that is the data most likely to be requested again
o Intellige nt algor ith ms de-s tage in a sequent ial ma nner
Understanding Flash Cell Endurance
Write cache management is essential to improving performance but it is also the key part of how VMAX All Flash helps extend the
endurance of flash drives. Flash drive longevity and endurance are most impacted by writes, particularly small block random writes.
Writing to a flash cell requires that the cell be first erased of any old data and then be programmed with the new data. This process is
called the Program and Erase Cycle (P/E Cycle). Each flash cell has a finite number of P/E Cycles that it can endure before it wears out
(can no longer hold data). Most modern flash cells can endure several thousand P/E Cycles.
One of the peculiarities of flash is that writes are s pread out acros s a flash page (typic al ly KBs in size); however, prior to the write
operation, the ex isting data in entire flash block (typically MBs in size) that the page is located in must be erased. Prior to erasing the
page, the flash controller chip finds an empty (erased) location on the drive and copies (writes) any existing data from the page to that
locati on. Because of how flash writes data, a simple 4 KB write from a host could result in many times that amount of data being written
internally on the drive causi ng P/E cycling on a large number of cells. Thi s write multiplying effect is called “Write Amplification” and is
detrimental to flash cell endurance. This effect is even more dramatic with small block random write workloads. In this situation, a large
number of small block random writes tends to “buckshot” acros s the drive, impacting an even greater number of cells and invoking P/E
cycling on a much larger cell area. Write amplification is not nearly as significant with larger sequential writes as this data is written
sequentially local to a single flash block, thereby aligning better with flash page sizes and containing the P/E cycling to a smaller area.
12
VMAX All Flash Write Amplification Reduction
Write Amplification must be properly controlled and mitigated in order to ensure the longevity of flash devices as uncontrolled write
amplification is the number one reason for premature wear out of flash storage. Controlling flash cell wri te amplification is one of the
VMAX All Flash’s greatest strengths and is what sets it truly apart from other competitor flas h arrays. Aside from intelligent caching
algorithms which keep data in cache as long as possible, the VMAX All Flash employs to additional methods to minimize the amount of
writes to flash. These methods are:
• Write Folding – Write Folding avoids unnecessary disk IOs when hosts re-write to a particular address range. This re-written
data is simpl y replaced in cache and never written to the flash drive. Write folding can reduc e writes to the flash drives by up to
50%.
• Write Coalescing – Write Coalescing merges subsequent small random writes from different times into one l arge sequential
write. These larger writes to the flash drives align much better with the page sizes within the flash drive itself. Using write
coalescing, VMAX All Flash can take a highly random write host IO workload and make it appear as a sequential write
workload to the flash driv es.
• Timefinder SnapVX Nocopy Linked Target Func tionality – Timefinder SnapVX provides very low impact (space effici ent) point
in time snapshots for source volumes. Typically when a user wishes to unlink a target volume from a snapshot, as often done
in setting up a development environment, the unlinking operation would require a full volume copy of the source volume to the
target in order for the target to be used after being unlinked. This would also result in a large i ncrease in back end capacity
usage with large amount of write operations on the back end drives. SnapVX eliminates this requirement as the point in time
image is still accessible after unlinking the nocopy target volume. This saves the back end flash devices from enduring a large
amount of write activity.
• Advanced Wear Analytics - VMAX Al l Flash also includ es advanced drive wear analytics optimized for high capacity fla sh
drives to make sure writes are distributed across the entire flash pool to balance the load and avoid exc es sive writes and wear
to particular drives. Not only does this help manage the flash drives in the storage pools, it makes it easy to add and rebalance
additional stor a ge into the sy st em.
All of the write amplification reduction techniques used by VMAX All Flash result in a significant reduction in writes to the back end,
which in turn significantly increases the longevity of the flash driv es used in the array .
Boosting Flash Perform ance with FlashBoost
DELL EMC is always looking to improve performance in its products. With every new hardware platform and release of software, the
company makes strong efforts to remove potential bottlenecks which can impede performance in any way. One feature that DELL EMC
introduced and has made standard as a part of HYPERMAX OS is “FlashBoost”. FlashBoost maximizes HYPERMAX OS efficiency by
servicing read requests directly from the back-end flash drives. This approach elimi nates steps required for processing IO through
global cache and reduces the latency for reads, particularly for flash drives. Customers with heavy read miss workloads residing on
flash can see up to 100% greater IOPS performance.
HYPERMAX OS
The VMAX All Flash engines leverage the trus ted and proven HYPERMAX OS. It combines proven industry-leading high av ailab ility ,
I/O management, quality of service, data integrity validation, data movement, and data security with an open application platform.
HYPERMAX OS features the first real-time, non-disruptive storage hypervisor that manages and protects embedded services by
extending high availability to services that traditionally would have run external to the array. The primary functions of HYPERMAX OS
are to manage the core operations performed on the array such as:
• Processing IO from hosts
• Implementing RAID protection
• Optimizing performance by allowing direc t access to hardware resources
• System Management and Monitoring
• Implementing data services including loc al and remote replication
VM AX ALL FLASH DATA SERVI CE S
The VMAX All Flash product line comes complete wi th best in class data services. In VMAX, data services are processes which help
protect, m anage, and move customer data on the array. These services run natively, embedded inside the VMAX itself using the
HYPERMAX OS hypervisor to provide a resource abstraction layer. This allows the data servi c es to share pooled resources (CPU
cores, cache, and bandwidth) within the array itself. Doing this optimizes performance across the entire system and also reduces
complexi ty in the environment as resources (system cache, CPU cores, and outside appliances) do not need to be dedicated.
Some of the most sought after data services that will be offered by the VMAX All Flash product line are:
• Remote Replication with SRDF
• Local Replication with TimeFinder SnapVX
• Embedded NAS (eNAS)
13
• eManagement – embedded Unisphere for VMAX
Remote Replication with SRDF
SRDF is perhaps the most popular data service in the enterprise data center as it is considered a gold standard for remote replication.
Up to 70% of fortune 500 companies running VMAX use this tool to replicate their critical data to geographically dispersed data centers
throughout the world. SRDF offers customers the ability replicate tens of thousands of volumes to a maximum of four different locations
globally.
VMAX All Flash runs an enhanced version of SRDF speci fic for all flash use cases. This version uses mul ti-core, mu lti-threading
techniques to boost performance; and powerful write folding algorithms to greatly reduce replication bandwidth requirements along
withsource and target array back end writes to flash.
There are three primary flavors of SRDF that a customer can choose from:
(1) SRDF Synchronous (SRDF/S) – SRDF/S delivers zero data loss remote mirroring between data centers separated by up to 60
miles (1 00 km) .
(2) SRDF Asynchronous (SRDF/A) – SRDF/A delivers asynchronous remote data replication between data c enters up to 8000
miles (12875 km) apart. SRDF/A can be used to support three or four site topologies as required by the world’s most mission
critical applications.
(3) SRDF/Metro – SRDF/Metro del ivers active-active high availability for non-stop data access and work load mobility within a
data center, or between data centers separated by up to 60 miles. SRDF/Metro allows for storage array clustering, enabling
even more resilienc y, agility, and data mobility. SRDF/Metro allows hosts or host clusters access to LUNs replicated between
two different sites. The hosts can see both views of the Metro Replicated LUN (R1 and R2), but it appears to the host OS as if
it were the same LUN. The host can then write to both the R1 and R2 devices simultaneously. Thi s use case accounts for
automated recovery and the seamless failover of applications thus avoiding recovery scenarios all together. Other key features
of SRDF Metro are:
• It provides concurrent ac c ess of LUNS /storage groups for non-stop data access and higher availability across metro
distances
• It delivers simpler and seamless data mobil ity
• It supports stretch clustering which is ideal for Microsoft and VMware environments
SRDF software is included in the VMAX All Flash FX software package, with no capacity based licensing. It can be ordered a la carte
as an addition to the F software package. Any hardware needed to support SRDF must be purchased separately.
Local Replication with TimeFinder SnapVX
Every VMAX All Flash array comes standard with the local repl ication data service Timefinder SnapVX as it is included as part of the F
package. SnapVX provides very low impact snapshots and clones for VMAX LUNs. SnapVX supports up to 256 snapshots per source
volume and up to 16 million snap shots per array. Users can assign names to identify their snapshots, and they have the option of
setting automatic expiration dates on each snapshot.
SnapVX provides the ability to manage consistent point-in-time copies for storage groups with a single operation. Up to 1024 target
volumes can be linked per source volume, providing read/write access as pointers or full-copy cl ones.
Local replication with SnapVX starts out as efficient as possible by creating a snapshot, a pointer based structure that preserves a point
in time view of a source volume. Snapshots do not require target volumes, share back-end allocations with the source volume and
other snapshots of the source volume, and only consume additional space when the source volume is changed. A single source
volume can have up to 256 snapshots.
Each snapshot is assigned a user-defined name and can optionally be assigned an expiration date, both of which can be modified later.
New management int er fa ces p rov ide the user with the ability to take a snapshot of an entire Storage Group with a single command.
A point-in-time snapshot can be accessed by linking it to a hos t accessible volume referred to as a target. The target volumes are
standard thi n LUNs. Up to 1024 target volum es can be linked to the snapshot(s) of a single s ource volume. This limit can be achieved
either by linking all 1024 target volumes to the sam e snapshot from the source volume, or by linking multiple target volumes to mul tiple
snapshots from the same source volume. However, a target volume may only be linked to a single snapshot at a time.
By default, targets are linked in a no-copy mode. This no-copy linked target functionality greatly reduces the amount of writes to the
back end flash drives as this eliminates the requirement of performing a full volume copy of the source volume during the unlink
operation i n order to use the target volume for host IO. This saves the back end flash devices from enduring a large amount of write
activity during the unlink operation, further reducing potential write amplification on the VMAX All Flash array.
Consolidation of Bloc k and File Storage Using eNAS
The Embedded NAS (eNAS) data service extends the value of VMAX All Flash to file storage by enabling customers to leverage vital
enterprise features including flash level performance for both block and file storage, as well as simplify management, and reduce
deployment costs by up to 33%. VMAX All Flash with the eNAS data servi ce becomes a unified block and file platfor m, usin g a multi -
controller, transaction NAS solution designed for customers requiring hyper consolidation for bl ock storage (the traditional VMAX use
case) combined with moderate capacity, high performance file storage in mission-critical environments. Common eNAS use cases
14
include running Oracle® on NFS, VMware® on NFS, Microsoft® SQL on SMB 3.0, home directories, and Windows server
consolidation.
Embedded NAS (eNAS) uses the hypervisor provided in HYPERMAX OS to create and run a set of virtual machi nes within the VMAX
All Flash array. These virtual machines host two m aj or elements of eNAS: software data movers and control stations. The embedded
data movers and control stations have access to shared system resourc e pools so that they can evenly consume VMAX All Flash
resources for both performance and capacity.
Aside from performance and consolidation, some of the benefits that VMAX All Flash with eNAS can provide to a customer are:
• Scalability – easily serve over 6000 active SMB connections
• Meta-data logging file system ideally suited for an all flash environment
• Built-in asynchronous file level remote replication with File Replicator
• Integration with SRDF/S
• Small attack surface – not vulnerable to viruses targeted at general purpose operating systems
The eNAS data service is included in the FX software package. It c an be ordered a la carte as an additional item with the F software
package. All hardware required to support eNAS on the VMAX All Flash must be purchased separately
Embedded Management (Embedded Unisphere for VMAX)
VMAX All Flash customers can take advantage of simplified array management using embedded Unisphere for VMAX. Unisphere for
VMAX is an intuitive man age m ent interf a ce that all ows IT managers to maximize hum an productivity by dramatically reducing the time
required to provision, manage, and monitor VMAX All Flash storage assets.
Embedded Unisphere enables customers to simplify management, reduce cost, and increase availability by running VMAX All Flash
management software directly on the array. Embedded management is configured in the factory to ensure mini mal setup time on site.
The feature runs as a container within the HYPERMAX OS hypervisor, eliminating the need for a customer to allocate their own
equipment to manage their arrays. Aside from Unisphere, other key elements of the eManagement data service include Solutions
Enabler, Database Storage Analyzer, and SMI-S management software.
Unisphere for VMAX delivers the simpli fication, flexibility, and automation that are key requirements to accelerate the transformation to
the all flash data center. For customers who frequently build up and tear down storage configurations, Unisphere for VMAX makes
reconfiguring the array even easier by reducing the number of steps required to delete and repurpose volumes. With VMAX All Flash,
storage provision to a host or virtual machine is performed with a simple four step process using the default Diamond class storage
service level . This ensures all applications will receive s ub-ms response times. Using Unisphere for VMA X, a customer can set up a
multi-site SRDF confi gurations in a matter of minutes.
Embedded Unisphere is a great way to manage a single VMAX All Flash array; however, for custom ers who require a higher pane of
glass view of their entire data center, DELL EMC offers Unisphere 360. Unisphere 360 software aggregates and monitors up to 200
VMAX All Flash / VMAX arrays across a single data center. This solution is a great option for customers running multiple VMAX All
Flash arrays with embedded management (eManagement) and who are looking for ways to facilitate better insights across their entire
data center. Unisp her e 360 pr ov ides stor age adm ini str a tors t he ability to view site-level health reports for every VMAX or coordinate
compliance to code levels and other infrastructure maintenance requirements . Customers can leverage the simplification of VMAX All
Flash management, now at data center scale.
Embedded Unisphere and Database Storage Analyzer are available with every VMAX All Flash array as they are included in the F
software package. Unisphere 360 is included in the FX software package, or can be ordered a la carte with the F software package.
Unisphere 360 does not run in an embedded environment and will require additional customer supplied server hardware.
VMAX All Flash – High Avai l abil it y and Res ilience
VMAX All Flash reliability, availability, and servic eability (RAS) make it the ideal platform for environments requiring always-on
availability. These arrays are architected to provide six-nines of availability in the most demanding, mission-cr iti cal envir o nmen ts. VM A X
All Flash availability, redundancy, and security features are listed below:
• No single points of failu r e—all components are fully redundant to withstand any component failure
• Completely re dund ant and hot -pluggable field-replaceable units (FRUs) to ensure repair without taking the s ystem offline
• Choice of RAI D5 or RA ID 6 deployment options to provi de the highest level of protection as desired
• Mi rrored cache, where the copies of cache entries are dis tributed to maximize availability
• HYPERMAX OS Flash Drive Endurance Monitoring – The nature of flash drives is that their NAND flash cells can be written to
a finite number of times. This is referred to as flash drive endurance and is reported by drive firmware as a “percentage of life
used”. HYPERMAX OS periodically collects and monitors this information and uses it to trigger alerts back to DELL EMC
Customer Support when a particular drive is nearing its end of useful life.
• Vault to flash with battery backup to allow for cache de-stage to flash and an orderly shutdown for data protection in the event
of a power failure
• Active-active remote replication via SRDF/Metro with read/write access to both Site A and Site B ensures instant data access
during a site failure.
15
• Fully non-disruptive upgrades, including loading HYPERMAX OS software from small updates to major releases
• Continuous system monitoring, call-home notification, and advanced remote diagnostics
• Data at Rest Encryption (D@RE) with integrated RSA® key manager, FIPS 140-2 compliant to meet stringent regulatory
requirements
• T10 DIF data coding, with extensions for protection against lost writes
• Detailed failure mode effects analysis (FMEA) during design of each component to ensure failure conditions can be handled
gracefully
• Extensive fault detection and isolation, allowing early wear-out detection and preventing the passing of bad data as good
• Service defined and scripted to ensure success, incl uding color-coded cabling, cable posi ti oning , scrip ted steps, and checks of
key parameters in those scripts
• All flash cache data vault capable of surviving two key failures, ensuring that the system comes back even when something
was broken before the vault and something else fails when returning from the power cycle
• Support for thermal exc ursions with graceful shutdown if, for example, a data center loses air conditioning
• Integrated data protection via DELL EMC ProtectPoint backup and rapid re store, combining the gold standards in
backup with industry leading SRDF replication technology
VM AX All Flash Confi gurations for Open Systems
For open system configurations, the VMAX Al l Flash brick is referred to as a “V-Brick”. Each initial open systems V-brick comes pre-
configured from DELL EMC manufacturing with its own system bay. Dual engine system bay configurations are exclusively used as
additional V-bricks are added to the system. Optionally, customer specific racks can be used as long as they are standard NEMA 19-
inch racks and meet DELL EMC standards for cable acces s and cool ing.
Open Systems V-Brick System Configurations for the VMAX 250F
All VMAX 250F V-Bricks include a base capacity of 11TBu. Capacity is delivered via flash drives sizes of 960GB, 1.92TB, 3.84TB,
7.68TB, and 15.36TB, and is upgradeable in increments of 11TBu Flash Capacity Packs. The VMAX 250F V-Brick engines contain 2
directors each with, 512GB, 1TB, or 2TB of memory and with dual 12 core processors per director. The VMAX 250F is shipped in a
dual engine cabinet configuration. A VMAX 250F system cabinet can scale out to accommodate a maximum of 2 full V-Bricks and 100
drives per floor til e yielding up to 1PBu in only 20U of rack space. The remaining rack s pace can be taken by an additional VMAX 250F
system or by customer hardw are such as serv ers and sw itch es.
The following diagram shows the VMAX 250F starter V-Brick system bay configuration along with a dual V-Brick system bay
configuration:
Figure 4. VMAX 250F Starter V-Brick Configuration and Dual V-Brick System Bay
Starter V-Brick
System Bay
V-Brick DAEs
2 x 25 slot, 2.5"
V-Brick Engine 10 U
Dual V-Brick
System Bay
V-Brick A
20 U
V-Brick B
System Power
Supplies
16
Open Systems V-Brick System Configurations for the VMAX 450F and VMAX 850F
All open syste m VMA X 450F / 850F configurat io ns are also built with V-Bricks, which include compute and a base capacity of 53TBu.
Capacity is deliver ed via fla sh driv es sizes of 960GB, 1.92TB and 3.84TB, and is upgradeabl e in increments of 13TBu Flash Capacity
Packs. The 450F and 850F Engines contain 2 directors each with 1TB or 2TB of memory and with dual processors per director (8 core
2.6GHz for the 450F and 12 Core 2.7GHz for the 850F) and ship in dual engine cabinets. A s ingle cabinet can accommodate 2 ful l V-
Bricks and 480 drives per floor tile with up to 1 useable PB per cabinet. The VMAX 450F scales up to 4 V-Bric ks and 2PB u per syste m
and the 850F up to 8 V-Bricks and 4PBu per system.
The following diagr am show s the VMAX 450F /850F starter V-Brick system bay configuration along with a dual V-Brick s y st em bay
configuration:
Figure 5. VMAX 450F / 850F Starter V-Brick Configuration and Dual V-Brick System Bay
The VMAX 450F model can scale out to four V-Bricks , which would require two systems bays (two floor tiles), while the VMAX 850F
can scale out to eight V-Bricks, which would require four system bays (four floor tiles). System bays can be separated by up to 25
meters u sing opt ic al connect ors.
Open Systems V-Brick Front End Connectivity Options
For V-Bricks, engine cooling fans and power supplies can be accessed from the front, while the I/O modules, management modules,
and control station can be accessed from the rear. Sinc e the number of universal I/O modules used in the V-Bric k engin e depen ds on
the customer’s required functionality, some slots can remain unused.
There are multiple supported V-Brick front-end connections which are available to support several protocols and s peeds. The table
below highlights the various front-end connectivity modules available to the VMAX All Flash V-Brick:
17
Table 5. VMAX All Flash Open S y stems V-Brick Engine Front End Connectivity Modules
Connectivity Type
Module Type
Number of Ports
Mix With Protocols
Supported Spee ds (Gbps)
Fibre Channel
8 Gbps FC
4
SRDF
2 / 4 / 8
Fibre Channel
16 Gbps FC
4
SRDF
2/ 8 / 16
SRDF
10 GigE
4
iSCSI
10
SRDF
GigE
2
None
1
iSCSI
10 GigE
4
SRDF
10
Cloud Array (CA)
8 Gbps FC
4
FC, SRDF
2 / 4 / 8
eNAS
10 GigE
2
None
10
eNAS
10 GigE (Copper)
2
None
10
eNAS Tape Backup
8 Gbps FC
4
None
2 / 4 / 8
The quantity of V-Bri ck front-end por t s scales, depending on protocol type, to a maximum of 32 for the VMAX 250F and 24 for the
VMAX 450F / 850F.
On the four port 8 Gbps and 16 Gbps Fibre Channel IO modules, a c ustomer can mix Fibre Channel host connectivity and SRDF using
different ports. This is also true for the four port 10 GigE IO module where host iSCSI connectivity and GigE SRDF can be intermixed
using different ports on the module. A customer can also mix Fibre Channel c onnectivity to the DELL EMC Cloud Array, along with host
Fibre Channel connectivity, and SRDF using the 8 Gbps Fibre Channel modules. The GigE IO modules set aside for eNAS are
dedicated and cannot be used for any other GigE connectivity type such as iSCSI and SRDF.
Other modules which are used by the V-Brick are shown in the table below:
Table 6. Other VMAX 250F / 450F / 850F Open Systems V-Brick Engine Modules
Module Type
Purpose
Vault to Flash
Flash for Vault and Metadata (4 x 800 GB for VMAX 450F /850F, 3 x 400 GB or 800
GB for VMAX 250F)
Internal Fabric
Internal InfiniBand Fabric Connections
Backend SAS
Backend SAS connection to DAEs (12 Gbps for VMAX 250F, 6 Gbps for VMAX
450F / 850F)
Compression
Adaptive Compression Engine (ACE) and SRDF Compression
The VMAX 250F will use up to 3 Vault to Flash modules while the VMAX 450F and VMAX 850F systems use up to 4 Vault to Flash
modules. The extra flash module required for the VMAX 450F and VMAX 850F syste ms is d ue to the larger usable capacities that
VMAX 450F and 850F can scale to. The vault to flash modules usually will occupy slots 0, 1, and 6 on the VMAX 250F V-Brick engine
while the vault to flash modules will usually occupy slots 0, 1, 6, and 7 on the VMAX 450F and VMAX 850F V-Brick engine.
The compression modules per f orm all opera tion s for the Adaptive Compression Engine (ACE) as well SRDF compressi on for t he
VMAX All Flash systems. Thi s resul ts in an offloading of the c ompre ss ion tas k from usi ng engine CPU core cycles. Each V-Brick will
use a pair of compression modules (one per each V-Bric k director). The compression modules are usually loca ted in slot 7 on the
VMAX 250F and slot 9 on the VMAX 450F / 850F.
The following figure depicts a typical layout for a VMAX 250F V-Brick engine:
18
Figure 6. Typical VMAX 250F V-Brick Engine Layout
Management
Modules
Vault
to Flash
Front-End
Connectivity
Back-End
Connectivity
Vault
to Flash
Front-End
Connectivity
Compression
Internal
Fabric
Slot
0
Slot
1
Slot
2
Slot
3
Slot
4
Slot
5
Slot
6
Slot
7
Slot
8
Slot
9
Slot
10
Odd
Director
Even
Director
Empty
Note: On the VMAX 250F, slot 5 is left empty (unus ed).
The following figure depicts a typical lay out for a VMAX 450F / 850F V-Brick engine:
Figure 7. Typical VMAX 450F / 850F V-Brick Engine Layout
Open System VMAX All Flash So f tware Packaging
In order to simplify the softw are order ing and management process, VMAX All Flash will offer two different software packages for the
VMAX 250F, VMAX 450F, a nd VMAX 850F in open systems environ ments.
The first option is known as the “F package” which can be considered a starter package. The F package includes HYPERMAX OS,
Embedded Management, SnapVX and an AppSync starter pack. Any software title supported on VMAX All Flash can be added to the
F package as an a la carte softwar e addition. The VMAX All Flash models whic h use the F package are referred to as the VMAX 250F,
VMAX 450F, and VMA X 850F.
The second option – the more encompa ss ing packa ge – is known as the “FX package”. The FX package includes everything that’s in
the F package as well a s SRDF /S, SRDF/A, SRDF/ STAR, SRDF/Metro, CloudArray Enabler, D@RE, eNAS, Unisphere 360 and ViPR
Suite. The FX is priced to offer a bundled discount over an equivalent F option which has many a la carte titles as additions. Customers
can add any title supported on VMAX All Flash to the FX in an a la carte fashi on as well, including ProtectPoint, the full AppSync suite,
and DELL EMC Storage Analytics. The VMAX All Flash models which use the FX package are referred to as the VMAX 250FX, VMAX
450FX, and VMAX 850FX. The following table details what software is included in each VMAX All Flash pac kage:
19
Table 7. VMAX All Flash Open Systems Software Packages
Feature
F Package
Included
F Package
a la Carte
FX Package
Included
FX Package
a la Carte
Notes
HYPERMAX OS
Includes Migration Tools, VVOLS, QoS
(3)
Embedded
Management
Includes Unisphere for VMAX, Database
Storage Analyzer, Solutions Enabler,
SMI-S
Local Replication
Includes Timefinder SnapVX
AppSync Starter Pack
Remote Replication
Suite (1)
Includes SRDF/S/A/STAR
SRDF/Metro (1)
Unisphere 360
Cloud Array Enabler (1)
D@RE (2)
eNAS (1) (2)
ViPR Suite
Includes ViPR Controller and ViPR SRM
ProtectPoint
PowerPath
AppSync Ful l Suite
DELL EMC Storage
Analytics
(1) FX Package i ncludes software licensing. Required hardware need to be order ed separ ate ly
(2) Factory conf igur ed. M ust be enable d dur ing the order ing pro ces s
(3) Includes host I/O limits
VMAX All Flash for Ma infra me
For mainframe configurations, the VMAX All Flash brick is referred to as a “zBrick”. Each initial mainframe zbrick comes pre-configured
from DELL EMC manufacturing with its own system bay. Dual engine system bay configurations are exclusively used as additional
zbricks are added to the system.
VMAX All Flash for mainframe is restricted to the VMAX 450F and 850F products. Both must run only 100% mainframe workloads as
no mixing of mainframe and open systems workloads is allowed. The VMAX 250F does not support mainframe workloads.
Mainframe zBrick System Configurations for the VMAX 450F / 850F
All mainframe VMA X 450F / 850F zBricks include a base capacity of 53TBu. Capacity is delivered via flash drives sizes of 960GB,
1.92TB and 3.84TB, and is upgradeable in increments of 13TBu zCapacity Packs. The 450F and 850F zBrick engin es contain 2
directors each with 1TB or 2TB of memory and with dual processors per director (8 core 2.6GHz for the 450F and 12 Core 2.7GHz for
the 850F) and shi p in dual eng ine cabinets. A single cab inet can accommodate two full zBricks and 480 dri ves per floor tile with up to
400 PBu per cabinet. The VMAX 450F scales up to four zBricks and 800TBu per system and the 850F up t o eight zBricks and 1.7PBu
per system. The mainframe zBrick does not support the Adaptive Compression Engine (ACE), ther e fore all sy ste m capa ci ties are
expressed in usable capacity.
20
Figure 8. VMAX 450F / 850F Starter zBrick Configuration and Dual zBrick System Bay
Starter zBrick
System Bay
zBrick Engine
zBrick 2 DAEs
(2.5" Flash Drives Only)
Dual zBrick
System Bay
zBrick A Engine
zBrick A DAEs
zBrick B Engine
zBrick B DAEs
The VMAX 450F model c an scale out to four zBricks, which would require two systems bays (two floor tiles), while the VMAX 850F can
scale out to eight zBricks, which would require four system bays (four floor tiles). System bays can be separated by up to 25 meters
using optical con nect ors.
Mainframe zBrick Front End Connectivity Options
For zBricks , engine cooling fans and power supplies can be accessed from the front, while the I/O modules, management modules, and
control station can be accessed from the rear. Sinc e the number of universal I/O modules used in the zBrick engine depends on the
customer’s required functionality, some slots can remain unused.
The zBrick supports FICON and SRDF front end connectivity. The table below highlights the various front-end connectiv ity modu les
available to the VMAX All Flash zBrick:
Table 8. VMAX All Fl ash Open Systems V-Brick Engine Front End Connectivity Modules
Connectivity Type
Module Type
Number of Ports
Mix With Protocols
Supported Spee ds (Gbps)
FICON
16 Gbps FICON
4
Single / Multi Mode
4 / 8 / 16
SRDF
16 Gbps Fibre
Channel
4
None
4 / 8 / 16
SRDF
8 Gbps Fibre
Channel
4
None
4 / 4 / 8
SRDF
10 GigE
4
None
10
SRDF
GigE
2
None
1
The quantity of zBrick front-end ports scales to a maximum of 32 when SRDF is not being used. When SRDF is used in the
configuration, one of the front end slots is taken by the SRDF compre ss ion module on eac h engine director. This limits the number of
available zBrick front end ports to 24. By default, each zBrick comes with two FICON modules.
21
Table 9. Other VMAX 450F / 850F Mainframe zBrick Engine Modules
Module Type
Purpose
Vault to Flash
Flash for Vault and Metadata (4 x 800 GB for VMAX 450F /850F)
Internal Fabric
Internal InfiniBand Fabric Connections
Backend SAS
Backend SAS connection to DAEs (6 Gbps for VMAX 450F / 850F)
Compression
(Optional)
SRDF compression only
The VMAX 450F and VMAX 850F systems use up to four Vault to Flash modules. The extra flash module required for the VMAX 450F
and VMAX 850F systems is due to the larger usable capacities that VMAX 450F and 850F can scale to. The vault to flash modules will
usually occupy slots 0, 1, 6, and 7 on the VMAX 450F and VMAX 850F zBrick engine.
The following figure depicts a typical VMAX 450F / 850F zBr ick engine which is co nfig ured for SRDF:
Figure 9. Typical VMAX 450F / 850F zBrick Engine Layout with SRDF
When SRDF is used in the configuration, each zBrick will use a pair of SRDF compression modules (one per each zBrick director). The
SRDF compression modules are usually located in slot 9 on the VMAX 450F / 850F. When SRDF is not used in the zBrick
configuration, a front end module can be placed into slot 9 providing additional front end connectivity. The following diagram shows a
typical non-SRDF zBrick engine configuration:
22
Figure 10. Typical VMAX 450F / 850F zBrick Engine Layout without SRDF
Management
Modules
Vault to Flash
Front-End
Connectivity
Back-End
Connectivity Vault to Flash
Front-End
Connectivity
Internal
Fabric
Slot
0
Slot
1
Slot
2
Slot
3
Slot
4
Slot
5
Slot
6
Slot
7
Slot
8
Slot
9
Slot
10
Odd
Director
Even
Director
VMAX All Flash for Mainframe Software Packaging
Software for mainframe support comes in two packages, z/F, the basic package and z/FX, a larger bundle of more advanced features.
Additional ly, many software features for mainframe can be ordered “a la Carte”. The packages are different to the standard all flash
packages and represent the core functionalities used by the mainframe customer. The following table highlights the VMAX All Flash for
Mainframe software packaging:
Table 10. VMAX Al l Flash for Mainfram e Software Pack ages
Feature
zF Package
Included
zF
Package a
la Carte
zFX Package
Included
zFX
Package a
la Carte
Notes
HYPERMAX OS
Includes Migration Tools, QoS
Embedded
Management
Includes Unisphere for VMAX, Databas e
Storage Analyzer, Solutions Enabler,
SMI-S
Local Replication
Includes Timefinder SnapVX
Mainframe Essentials
Remote Replication
Suite (1)(3)
Includes SRDF/S/A/STAR
Unisphere 360
AutoSwap
D@RE (2)
zDP
GDDR (3)
(1) zFX Package includes software licensing. Any additional required hardware needs to be ordered separately
(2) Factory conf igur ed. M ust be enable d dur ing the order ing pro ces s
(3) Use of SRDF/STAR for mainframe requires GDDR
23
SUMMARY
VMAX All Flash is a ground breaking all flash array designed for the most demanding and critical workloads in the enterpris e data
center. Its unique modular architecture allows it to massively scale while delivering predictable high performance regardles s of the
workload. Built into the array are complex algorithms which maximize flash perform ance, while greatly enhancing flash drive enduranc e.
Its unique data services and highly available 6x9s architecture make it a premier choice for the enterprise environment where ease-of-
use coupled with trusted dependability is an essential requirement.
REFERENCES
DELL EMC VMAX Local Replication Technical Note – P/N H13697
DELL EMC VMAX Unified Embedded NAS Technical Note – P/N H13904
DELL EMC VMAX Reli ability, Availability, and Serviceability Technical Note – P/N H13807
DELL EMC VMAX SRDF/Metro Overview and Best Practices Technical Note – P/N H14556
DELL EMC VMAX3 and VMAX All Flash Quality of Service Controls for Mul ti-Tennant Environments