a flashy data center starts with storage · ibm power systems ibm flashcore™ technology optimized...

Dave Gimpl, IBM

A Flashy Data Center Starts With Storage

IBM's Flash (and other) StorageiCommunity May 9th 2019

2019 IBM Systems Technical University

2

2 © Copyright IBM Corporation 2019

2019 iCommunity – IBM Storage

Please note

— IBM’s statements regarding its plans, directions, and intent are subject to change or withdrawal without notice and at IBM’s sole discretion.

— Information regarding potential future products is intended to outline our general product direction and it should not be relied on in making a purchasing decision.

— The information mentioned regarding potential future products is not a commitment, promise, or legal obligation to deliver any material, code or functionality. Information about potential future products may not be incorporated into any contract.

— The development, release, and timing of any future features or functionality described for our products remains at our sole discretion.

— Performance is based on measurements and projections using standard IBM benchmarks in a controlled environment. The actual throughput or performance that any user will experience will vary depending upon many factors, including considerations such as the amount of multiprogramming in the user’s job stream, the I/O configuration, the storage configuration, and the workload processed. Therefore, no assurance can be given that an individual user will achieve results similar to those stated here.


Agenda

— History

• FlashSystem

• Storwize, Spectrum Virtualize

— Current Offerings (& recent deliverables since we last met)

• FlashSystem

• Storwize

• NPIV*

• FlashCore Forever

• OOPS

• NVMe (both kinds)

— Announced/Upcoming Technology

— Futures

— Misc

— Q&A/Discussion

• Be interrupt driven (storage humor) and interactive (database humor)



History


2019 IBM Systems Technical University 6

IBM Systems Flash Storage Offerings Portfolio

6© Copyright IBM Corporation 2017

IBM ElasticStorage Server

DS888xF

Business Critical

z/OSPower HAPower i HA

• Extreme performance• Targeting database acceleration

& Spectrum Storage booster• Inline hardware compression• 3D TLC Flash

FlashSystem 900 Application acceleration

IBM FlashCore™ Technology Optimized

FlashSystem A9000

FlashSystem A9000R

• Full time data reduction

• Workloads: Cloud, VDI, Vmware

• 3D TLC Flash

Large

deployments

FlashSystem V9000

Highest

PerformanceCloud service

providers


• Workloads: Mixed and cloud

• 3D TLC Flash

StorwizeV7000F

Storwize V5030F

Entry Storage

Virtualization

Midrange Storage

Virtualization

Enhanced data storage functions, economics and flexibility with sophisticated virtualization

SVC

• Simplified management• Flexible consumption model

• Virtualized, enterprise-class, flash-optimized, modular storage• Enterprise class heterogeneous data services and selectable data reduction

• Business critical, deepest integration with z Systems

• Superior performance & reliability

• Three-site / Four-site replication

• DS8884F, DS8886F, DS8888F

IBMPowerSystems

Big Data

• Consolidates file & object workloads

• Faster data analysis

• Global sharing

Flexible Converged Infrastructure with power of Storwize V5030F and Spectrum Scale

• 3D TLC Flash

BEFORE


Enhanced data storage functions,

economics and flexibility with

sophisticated virtualization

IBM Systems Flash Storage Offerings Portfolio

IBM Elastic Storage Server

DS888xF

Business Critical

• z/OS / AIX• Power HA• Power i HA

• Business critical, deepest integration with z Systems

• Superior performance and reliability

• Three-site / Four-site replication

• DS8884F, DS8886F, DS8888F

Big Data

• Consolidate file & object workloads

• Faster data analysis• Global sharing

• Extreme performance• Targeting database

acceleration & Spectrum Storage booster

FlashSystem 900 Application acceleration

FlashSystem A9000

FlashSystem A9000R


• Workloads: Cloud, VDI, VMware

Large deployments

FlashSystem 9100Models 9110, 9150

NVMe accelerated Virtualizing the DC

Cloud service

providers


• Workloads: Mixed and cloud

StorwizeV5030F / V7000F

Entry /

Mid-Range

SVC

Simplified managementFlexible consumption model

Virtualized, enterprise-class, flash-optimized, modular storageEnterprise class heterogeneous data services and selectable data reduction

Flash Core ModuleSuperior endurance and better performance

• FIPS 140-2• Compression

IBM PowerSystems

IBM FlashCore™ Technology Optimized

AFTER


FlashCorePortfolio Integration

IBM Tech U © 2019 IBM Corporation

2019 IBM Systems Technical University© Copyright IBM Corporation 2018

IBM Storage and SDI

Application Acceleration

FlashSystem 900

Scale-Out FileScale-Out BlockScale-Out Object Virtualized Block Traditional NAS/File


IBM Storage and SDI


FlashSystem 900

Enterprise NVMe Flash Array

FlashSystem 9100& Storwize V7000



IBM Storage and SDI


FlashSystem 900



FlashSystem A9000R



IBM Storage and SDI


FlashSystem 900



IBM Integrated Analytics System

FlashSystem A9000R


2019 IBM Systems Technical University© IBM Corporation 2018 13

Designed for NVMe

Transformation of performance and capacity density

Up to 10 Million IOPS in 8u

Up to 2PB of data storage in 2U• Assumes 5:1 Data reduction

Performance and density to Modernize and Transform

What fast really is …IBM FlashSystem 9110 and 9150


Why IBM Storage for the enterprise?

Flash Software Defined Infrastructure

Data Management and Protection

• All Flash solutions for every workload

• High performance as well as high operational and agility benefits

• Next Generation Flash and Storage Class memory enables cognitive infrastructures

• Spectrum Suite provides single $/TB access to the broadest portfolio of SDS (Block, File, Object, Global File system, Backup and management)

• Software defined storage, compute grid workload management and API driven automation

• #1 in SDS (IDC)

• Protect, Recovery, Archive, Manage/Leverage Data Copy and overall Storage Management

• Cognitive management and predictive support

• #1 in Storage Software (IDC) and #2 in Data Protection

Help clients simultaneously modernize their traditional application infrastructure and transform their businesses with new workloads

True Private Cloud (Hybrid Cloud) High Performance Data Analytics (HPC/HPDA) Cognitive Data Management


Current Offerings (& recent deliverables since we last met)

— FlashSystem 900 Model AE3 does inline hardware compression

— FlashSystem 900:

• XL module

• NVMe fabric support



Compression AlgorithmInline, At Speed, Hardware Compression

The FlashSystem compression algorithm is a Modified Dynamic GZIP algorithm

• Always on because there is NO performance penalty, even for incompressible data.

• Combines LZ1 with a form of Pseudo Dynamic Huffman.

• Implemented completely in hardware - no processor intervention. This is not just a software implementation with hardware offload!

• This technology originated with Z and we have adopted it to fit into a flash controller.

• Takes advantage of already existing LSA (Logical Structured Array) mapping.


Compression PerformanceInline, At Speed, Hardware Compression

Compression and decompression is performed on individual logical pages inside each Flash Module

• First step in the inbound data path, before any logical-to-physical mapping occurs

• Less data to transfer in backend making up for small added latency (7uS)

• Incompressible data only consumes 1:1 (e.g., no meta-data bloat due to existing LSA re-use)

Compression and Decompression completely transparent above the Flash Module except for space management. Performance scales linearly with number of engines.

• Each Flash Module has 1 or 2 engines, depending on the type of Module


Compression Data ProtectionInline, At Speed, Hardware Compression

Decompression is performed as the last step in the outbound data path

• Directly prior to returning the requested data.

• No Store and forward

Decompression is performed as a check step in the inbound data path. After incoming data is compressed, it is run through the decompressor real-time and checked against original data before committing

Data protection (ECC) is implemented on top of compressed data.

• Allows garbage collection and other background data transactions to operate on compressed data


Compression Engine Block DiagramInline, At Speed, Hardware Compression

Compressor

LZ Encoding

Build Huffman Tree

HuffmanEncoding

Uncompressed Input Data

Compressed Data Header

A header is added to the compressed data block containing a description of the Huffman tree that was used to encode the data

1 Page e.g. 4KB

Decompressor

Compare

Flash Module

Flash Chips


FlashSystem 900 Performance

— World Class Performance even with 3DTLC

—The same or better performance then FS900 AE2!—Response time for real workloads slightly better with

compressible data then FS900 AE2.—Tuning Free, Consistent and predictable

20

FlashSystem 900 4K Random Write (Uncompressible Data vs. Comp Ratio=2):

AE2 12x2.6TiB, 90% Physical vs AE3 12x8.9TiB, 80% PhysicalFlashsystem 900 AE3: 4KB Random Mix (Comp Ratio=2):

12x8.9TiB, 80% Physical Capacity Used


Client Results using FS900 inline hardware compression

21 © 2018 IBM Corporation

5.0 GB/s

Bandwidth 5.0 GB/s

3.5 GB/s

115K IOPS

IOPS 115K

80K IOPS

1.6ms

Read Latency 1.6ms

6.7ms

.7ms

Write Latency .7ms

8.5ms

SVC w/ AE3 Compression

Real-time Compression

43%increasedbandwidth

44%increasedIOPS

4Xlower readlatency

12Xlower writelatency


FlashSystem 900 AE3 Configuration Matrix

22

• Additionally:• No mixing of MicroLatency modules of varying capacity• FlashSystem 900 chassis does not support mixing MLC & TLC flash media• Systems cannot be upgraded in the field to new flash media (MLC TLC)• Large capacity modules have the same effective capacity as the Medium modules. These are designed for

system that need additional capacity but have mixed workloads that may not be able to take full advantage of compression.

ModuleSmall (3.6TBu) Medium (8.5TBu) Large (18TBu) XLarge (18TBu)

Min Max Min Max Min Max Min Max

Qty 6 8 10 12 8 10 12 8 10 12 8 10 12

Usable 14.4 21.6 28.8 36.1 51.3 68.4 85.5 108 144 180 108 144 180

Expected Effective Cap. (2:1 typical)

28.8 43.3 57.7 72.2 102.6 136.8 171 NA NA NA 216 288 360

Max Effective 43.9 65.9 87.9 109.9 131.9 175.9 219.9 131.9 175.9 219.9 263.8 351.8 439.8

Foruncompressible

workloads New Offering


FlashSystem 900 Technical Foundation

FlashSystem 900(9840-AE2, 9843-AE2)

Next gen. FlashSystem 900(9840-AE3, 9843-AE3)

FlashSystem 900 AE3December Updates

Flash media MLC 3D TLC 3D TLC

Capacity points 13 10 13

Minimum Usable Capacity (TBu) 2.2 14.4 (3D TLC) 14.4

Maximum Effective Capacity (TBe) 57 220 440

Native compression No Yes Yes

RAID RAID 5 + Hot spare RAID 5 + Hot Spare RAID 5 + Hot Spare

Latency: write 90µs 95µs 95µs

Latency: read 155µs 155µs 155µs

IOPS: read (100%, random) 1,100,000 1,100,000 1,100,000

IOPS: read/write (70%/30%, random) 800,000 900,000 900,000

IOPS: write (100%, random) 600,000 600,000 600,000

Bandwidth: read (100%, sequential) 10 GB/s 10 GB/s 10 GB/s

Bandwidth: write (100%, sequential) 4.5 GB/s 4.5 GB/s 4.5 GB/s

Encryption Local key management Local key management, SKLM Local key management, SKLM

Interfaces 8 x 16Gb FC16 x 8Gb FC8 x 40Gb QDR InfiniBand

8 x 16Gb FC16 x 8Gb FC8 x 40Gb QDR InfiniBand

16 x 16Gb/8Gb FC / NVMe-FC8 x 40Gb QDR InfiniBand

Remote support No Yes Yes

Warranty 1 year or 3 year 1 year or 3 year 1 year or 3 year

23


New XL Flash Module

— Same 18TB Usable capacity as current Large Capacity Module

— Added additional DRAM to increase the effective capacity limits

— Effective capacity limit now supportable up to a maximum of 43.98TBe

• This maximum is depended on the data that is stored on the module.

• If the data is compressible to at least 2.44:1, then the full effective capacity can be used on each module

— If uncompressible data is stored on the module, then the module can store up to 18TBu; same as the current Large Module

— Can be MES on older AE3 systems

• Physical module is different, so new modules and data movement required


New 4x16Gb FC-NVMe Adapter

— Supported up to 4x16Gb per Adapter

• Will down train to 8Gb and 4Gb

• Simplified configuration and optics to a single option

— Supported NVMe-FC and FC on the same ports

— Can be MES for AE3 systems that have the current FC adapter

• Must have 1.6.x firmware to be able to use the adapter / NVMe-FC


NPIV

— Technique in storage system to “virtualize” its own WWPNs when presented to host, so that during upgrades and other maintenance procedures the hosts multipathing never knows it

— (industry slang term) “No loss of light”

• Derived from what looks like the fiberoptic (laser) connections never go away

— First introduced over 2.5 years ago in FlashSystem, Spectrum Virtualize in 7.7

— No reasons (nor excuses) of moving forward

• 7.6 support ended April 30th, 2019

• 7.7 is now the oldest release, and it has it

• “least common denominator” factor: everybody can use it now

— Benefits:

• Servers never know when storage is upgrading firmware/software

• Spare Node provides even more availability and performance protection, chiefly leveraged from NPIV

— Barriers to Adoption

• Does involve SAN zoning changes

• AIX LUN setting of “dynamic tracking” is requiredo AIX defaults this to off, needs to be on



FlashCore Forever

— Started Nov 24 2015

— FlashSystem V9000, 9848 (3 year machine type)

— Once the controllers reach 3 years old, customer is entitled to receive a FS9150 (empty, no NVMedrives) no HW purchase price, just have to purchase HWMA on it

• Sub

— Allows for re-use of the AE2 (and AE3) flash behind the FS9150 for no additional (External Virtualization) licenses

— Allows you to keep current on the Intel processors in the controller for no additional HW cost

— A nice performance boost too

• AC2 controllers only had 16 cores, FS9150 have 26

— Nice/graceful segway from V9000 to FS9150

— Worth consideration

• Re-use, re-purpose

• Product continuity



OOPS

— An exclamation when an accident is made

— #2 on the Top 10 popular Britney Spears

— Out Of Physical Space

• This is the correct response, is something to exclaim/shout (4 letter) words about when it occurs

• Is nothing to sing about

— Why is this happening now?

• Disk slow down as they fill up

• Flash doesn’t

• Operations notices a performance issue (on disk systems) then learns its really running low on free space

— Good reading material, for BP and customers (and IBMers) alike

• https://www-01.ibm.com/support/docview.wss?uid=ibm10735459



Preventing:Redo Capacity Plan using

actual CR

Planning:Capacity Planning for

Compression

Do you know your data

compression ratio?

Yes

No

Can you run Comprestimator

?

YesRun

Comprestimator to get

compression ratio (CR)

No

Compression ratio (CR)

Compression ratio is 1:1

(CR=1)

Physical Capacity = ((needed Capacity) / CR ) / .85

Planned Physical Capacity

Monitoring:Monitor Compression Ratio

Is the FS900

actual CR matching

the planned CR?

Have you hit OOPS

yet?

No

Yes

No

Recovery:Call IBM Support

Yes


NVMe

As Hardware

— Is a Device Specification / form factor / electronics specification

— Keep in mind both dimensions of its device composition

• Treats flash like memory – low latencyo An Informed Consumer will…then look for latency to be in the double digit mircoseconds (e.g. 50)

o A previously Informed Consumer…already knows FlashSystem is already transacting in this realm

• Sits on a PCI bus, typically PCIe gen4o An Informed Consumer will…then look for at least double the bandwidth

As an API, software, device driver, access protocol

— Largest benefit of NVMe is modern, memory like API set

• Half the commands/APIs as that of SCSI (which was invented for spinning HDD)o Less commands means less device driver code to process, which means less host CPU drain

o An Informed Consumer will…then look for observable drop in CPU for processing same workload, or, allow same amount of CPU to process more IO in any given time frame

• Massive Queue/BWo SCSI had (typically) 32 queue depth, and typically 2 queues

o NVMe has 65 thousand queue depth, and 65 thousand queues

o An Informed Consumer will…then look for differentiation on bandwidth30 © Copyright IBM Corporation 2018

2019 iCommunity – IBM Storage© IBM Corporation 2018 31

Drive greater productivity and faster reporting withIBM FlashCore Modules

— Built in, performance neutral hardware compression andencryption

— Using next gen 64 layer 3DTLC in just 9 months

— Outstanding data reliability

— Cognitive Algorithms for Wear Levelling, Health binning, Heat segregation and media management

— Intelligent media management that intelligently keeps settings ideal to prevent inconsistent performance.

—Endurance without latency penalty

—FIPS 140 certification

4.8TB, 9.6TB, 19.2TB capacity options with up to 3:1 compression

Agile

IBM FlashCoreTM technology delivers key differentiators


• Developed our own characterization platform

• Tested thousands of flash blocks

• Endurance not just measured – Techniques developed to extend endurance

• Experimented with different “knobs” Micron provided

• Experimented with application use models to reduce write amplification in real life use cases.

• Elite team of IBM researchers, semiconductor and flash experts from acquisition and other locations

IBM’s World Class Enterprise Flash Expertise

Result is breakthrough technology providing worry free usageeven in write intensive environments!


What is NVMe? (Non-Volatile Memory Express)

• NVMe is a communications interface and protocol

• NVMe is designed from the ground up to deliver high bandwidth and low latency storage access

• NVM Express defines an efficient interface for host software to communicate with a non-volatile memory subsystem over PCI Express (NVMe over PCIe)

• Functionally analogous to SAS and SATA but designed to reduce overhead from drivers, OS and application

• Leverages poll-based I/O completions as opposed to interrupt-based completions

• NVMe uses PCIe fabric

• Multiple devices today in the market

• Multiple form factors including 2.5” drives

• NVMe has been designed for high performance

• Increased IOPs, bandwidth and lower latency

• Exploit Flash and next-generation Non-Volatile Memories

• Leverage multicore environments, high I/O parallelism

33


Native NVMe Multipath with ANA

• NVMe native multipath is a different implementation than the Linux Device Mapper

• The Linux Device Mapper is a generic module that supports Multipath for different types of devices, whereas NVMe native multipath is tailored for NVMe only

• Native multipath is the only one that supports ANA which is the only NVMe path prioritizer. Thus it is required to use Native multipath and not Device mapper when applicable

• It is supported on Linux SUSE

• On SLES15 – it is enabled by default

• On SLES12SP4 – it is disabled by default, need to turn it on

• On SLES12SP3 – not supported

• The performance is much better with NVMe native multipath

• In Native multipath every Namespace has one instance

34


The Value of NVMe / NVMe-oF

35

Within Storage Controller

End-to-End NVMe Protocol

MediaFlash Drive

Host OS Application

User Space

MediaFlash Cache

Controller to Host

Connection Technologies

NVMe-oF protocolFor Host Connection

NVMe protocolFor Flash Storage Connection

• Optimized for Flash

• Flash Media with different formats

• Fast and getting faster

• Reduce host application license costs

• Future proof investment

• NVMe end-to-end strategy

PersistentCache


FS9100 SCSI vs NVMe/FC: Typical 70% READ workload

— Workload

• IO size 4KB @70% Read / 30% Write cache hit

• Identical workload of 200K IOPs on SCSI and NVMe/FC

• 32 SCSI devices with total QD=32

• 32 NVMe/FC devices with 4 associations and total 64 queues

Performance

• SCSI and NVMe/FC IOs show identical response time

• SCSI performance max out at QD32

• NVMe/FC delivers same IOPs at half CPU consumption

• Code efficiency in NVMe/FC host stack

• Potential queuing in NVMe/FC gives same response time

with lesser CPU cost than SCSI

36


FS9100 SCSI Vs NVMe/FC: On IO intensive 70% READ workload

— Workload

• IO size 4KB @70% Read / 30% Write cache hit

• Maximum workload with QD 512

• 32 SCSI devices

• 32 NVMe/FC devices with 4 associations and total 64 queues

Performance

• NVMe/FC IOPs scale to 400K IOPs

• NVMe/FC show 50% latency drop over SCSI

• NVMe/FC IOPs limited by Storage target port capability

• SCSI IOPs limited to 220K IOPs

• SCSI performance limited by host stack bottleneck

• SCSI drives CPU usage almost to 70%

37


FS9100 70% READ: SCSI Vs NVMe/FC

SCSI Vs NVMe/FC

• Exhibit identical IOPs, Response time till 270K IOPs.SCSI hits hockey stick curve @270K IOPs,NVMe/FC perf peaks at 400K IOPs

• Sub-millisec Response time up to 512 QD for NVMe/FC

• Exhibit similar Response time up to 512 QD

• Less room for application after SCSI jumps CPU usage with higher QD

• Fairly low CPU usage with high NVMe/FC IOPs

Workload: IO size 4KB @70% Read, 30% Write 32 devices, total 64 NVMe queues, varying QD

38


IBM Storage Portfolio and NVMe / NVMeoF Strategy

• NVMe Use Cases

• Use NVMe drives as a caching medium, either on the host or in the storage system, to reduce the read latency for cache-friendly workloads• Local read-only cache (LROC) function of IBM Spectrum Scale is already capable of using NVMe

drives in this way

• Integrate both commodity NVMe and IBM FlashCore media• FS9100, V7000G3, V5100/F available

• Continue to take advantage of NVMe over Fabrics as a protocol for host attachment• Spectrum Virtualize family (SVC, V7000. V5100/F) NVMe/FC available

• FlashSystem family (FS9100, V9000, ) NVMe/FC available

• Spectrum Virtualize family (SVC, FS9100, V9000, V7000) Ethernet adapters are RoCE/iWARP ready

• A9000/A9000R FC adapters are NVMe/FC ready

• DS8000 already leverages zHyperLink for low latency on IBM Z

• Use NVMe over Fabrics as a protocol for back-end storage access• Similar approach is already used today by IBM Spectrum Accelerate, which uses the user-space I/O

paradigm to access back-end storage over RDMA in the FlashSystem A9000 and A9000R39


Announced/Upcoming Technology

— FlashSystem 900

• 1.6.1.0

— Spectrum Virtualize 8.3

• Support for latest NVMe capable device: Storwize V5100

• DRP performance improvements

• Object based access control (nice for multi-tenant, multi-operator, e.g. cloud)



Additional Updated FlashSystem 900 Features

— Faster rebuilds (Next Release)

• If a module is not 100% consumed, rebuilds will no longer rebuild unallocated pages.

— Update GUI

• New System View and Update Performance Panels

— Email Notification Emergency Power Loss

• If email notification is setup, on EPL, the system will send a notification to the configured email address that the enclosure has lost power

— Improved capacity reporting for usable and effective capacity

• CLI will now allow the reporting of effective, usable, and used capacity on each flash module

— DNS-enabled LDAP server entries


FS900 R3 GUI Enhancement Overview

— New Graphical User Interface

— Provides a new Dashboard view of the system• At a glance overview of performance, capacity

and system health

— Enhancements for use with mobile devices• Including Event Flag based performance charts

— Performance graphs overlaid with events

— Improvements to “strongly encourage” enabling of Call Home and Remote Access

— “Open PMR” button

— Capacity over time GUI• Will now give clients more insight on how they

are using their capacity over time

— Support Compression additions

42


Futures

— What would the next gen FlashSystem 900 model AE3 look like?

• Could it be an AE4? What capacity might it have?

— Protocol Focus

— HBA Focus

• What would make 32Gb fibrechannel interesting?



Battery Module2x Host SwappablePower Monitoring

Flash Modules12x Host SwappableOne or two PCBAs, multiple capacity points

FS900 AE3 Expanded View Front

Enclosure LEDsEnclosure Role UpOf system status

Front BezelRemovable Bezel to enableOEM and integrated configuration

Slot FillersSlot fillers used to control airImpendence and provide cleanair to canisters

44


Misc

— Best Practices of flash

• 85%o For sunny days, and rainy days


2019 iCommunity – IBM Storage 48© Copyright IBM Corporation 2017

IBM HyperSwap® Enduring Economics

• Provides highly available, concurrent access to a single copy of data from data centers up to 300 km (186 miles) apart

• Enables non-disruptive storage and virtual machine mobility between data centers

• Works with standard multipathing drivers

• Groups multiple volumes together for high availability

High Availability Clusters IBM availability guarantee


Q&A



Thank you

Dave GimplFlashSystemIntegration ArchitectChief Architect, FlashSystem [email protected]+1-507-269-4081



Notices and disclaimers

— © 2019 International Business Machines Corporation. No part of this document may be reproduced or transmitted in any form without written permission from IBM.

— U.S. Government Users Restricted Rights — use, duplication or disclosure restricted by GSA ADP Schedule Contract with IBM.

— Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors. IBM shall have no responsibility to update this information. This document is distributed “as is” without any warranty, either express or implied. In no event, shall IBM be liable for any damage arising from the use of this information, including but not limited to, loss of data, business interruption, loss of profit or loss of opportunity. IBM products and services are warranted per the terms and conditions of the agreements under which they are provided.

— IBM products are manufactured from new parts or new and used parts. In some cases, a product may not be new and may have been previously installed. Regardless, our warranty terms apply.”

— Any statements regarding IBM's future direction, intent or product plans are subject to change or withdrawal without notice.

— Performance data contained herein was generally obtained in a controlled, isolated environments. Customer examples are presented as illustrations of how those

— customers have used IBM products and the results they may have achieved. Actual performance, cost, savings or other results in other operating environments may vary.

— References in this document to IBM products, programs, or services does not imply that IBM intends to make such products, programs or services available in all countries in which IBM operates or does business.

— Workshops, sessions and associated materials may have been prepared by independent session speakers, and do not necessarily reflect the views of IBM. All materials and discussions are provided for informational purposes only, and are neither intended to, nor shall constitute legal or other guidance or advice to any individual participant or their specific situation.

— It is the customer’s responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customer’s business and any actions the customer may need to take to comply with such laws. IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer follows any law.

51


Notices and disclaimers continued

— Information concerning non-IBM products was obtained from the suppliers of those products, their published announcements or other publicly available sources. IBM has not tested those products about this publication and cannot confirm the accuracy of performance, compatibility or any other claims related to non-IBM products. Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products. IBM does not warrant the quality of any third-party products, or the ability of any such third-party products to interoperate with IBM’s products. IBM expressly disclaims all warranties, expressed or implied, including but not limited to, the implied warranties of merchantability and fitness for a purpose.

— The provision of the information contained herein is not intended to, and does not, grant any right or license under any IBM patents, copyrights, trademarks or other intellectual property right.

— IBM, the IBM logo, ibm.com and [names of other referenced IBM products and services used in the presentation] are trademarks of International Business Machines Corporation, registered in many jurisdictions worldwide. Other product and service names might be trademarks of IBM or other companies. A current list of IBM trademarks is available on the Web at "Copyright and trademark information" at: www.ibm.com/legal/copytrade.shtml

52

a flashy data center starts with storage · ibm power systems ibm flashcore™ technology optimized...

Documents