flash technology for all
TRANSCRIPT
Why the Time is Now to Transform Your Business with Flash
Servers and networks get relentlessly
faster and cheaper with Moore’s Law, and
as a result, are now thousands of times
faster than the mechanical disk storage to
which they connect. From the perspective
of today’s processor, the fastest hard drive
appears slower than tape did twenty years
ago. As a result of this growing mismatch,
much of consumer technology has already
transitioned from disk to flash memory: flash
powers our smartphones and tablets, and
underlies high performance websites such as
Google, Facebook, and Apple.
Why then is business still shackled to
mechanical disk? Until recently, flash has
been substantially more expensive than
disk, and largely incompatible with existing
enterprise computing architectures. Pure
Storage was founded to overcome these
hurdles by delivering all flash storage that
preserves investment in applications and
infrastructure, and costs less than the
mechanical alternatives. Accomplishing
this required a holistic rethink of enterprise
storage, and yielded a product radically better
in every dimension: 10X faster, 10X more
power & space efficient, 10X simpler, and
more reliable, but who’s CapEx is comparable
to the disk alternatives, and who’s OpEx is
radically better.
3
All-Flash Arrays: The Evolution of Storage & Flash
5
Transitioning to the All-Flash Enterprise
6
Solving the Flash for All Economics
7
Flash for All Applications
8
The Business Beneits of Flash for All
9 Flash for All: It’s Time
10 From the Gartner Files: Cool Vendors in Storage Technologies, 2013
15 About Pure Storage
Featuring research from
Flash for AllHow every business can improve eficiency, compete better, and
deliver compelling user experiences with lash
Scott Dietzen
CEO
2
Transitioning to an all-flash enterprise enables three important
business benefits:
1. Achieve business results that are not even imaginable with
disk. Legacy technologies introduce bottlenecks in the compute
architecture, but their real bottleneck is they constrain the
imagination of your business architects. They constrain what you
think is possible: how much data you can analyze, how fast you can
respond to a customer, how quickly you can launch a new service?
Moving to flash can speed your existing revenue-driving business
applications, but it will also unlock a whole new level of exploration
by your innovators into the art of the possible. What will your
business achieve with flash?
2. Deliver differentiated, compelling end-user experiences.
Computing and IT is fundamentally a tool in service to an ultimate
end-user, whether that end-user be a consumer using a mobile
device, a trader consuming real-time financial analysis, or a
doctor logging patient information into a VDI terminal. Computing
is about enabling people to feel, enjoy, and achieve. Whatever
your business, you compute to deliver compelling end-user
experiences, and flash is key to taking those experiences to the
next level.
3. Achieve new levels of efficiency in the storage factory. As
enterprise IT evolves to private cloud architectures, it is increasingly
clear that storage is a key platform service in the private
cloud. Moving to flash is a critical initiative to achieve more
efficient storage service delivery. Flash combines fundamental
improvements (10X less power, 10X less rack space, 10X less
power/cooling) with a new level of administrative simplicity that
allows for a dramatically less expensive storage factory.
Pure Storage has cracked the code on how to deliver an all-flash
storage platform that enables flash for all: flash for all applications,
flash for all customer types/sizes, and flash for all budgets. But the
real story is you: what you will achieve by going all flash. We hope
this paper will help set you down the path of discovering your all-flash
future.
Source: Pure Storage
Flash for All is published by Pure Storage. Editorial content supplied by Pure Storage is independent of Gartner analysis. All Gartner research is used with Gartner’s permission, and was originally published as part
of Gartner’s syndicated research service available to all entitled Gartner clients. © 2013 Gartner, Inc. and/or its affiliates. All rights reserved. The use of Gartner research in this publication does not indicate Gartner’s
endorsement of Pure Storage’s products and/or strategies. Reproduction or distribution of this publication in any form without Gartner’s prior written permission is forbidden. The information contained herein has been
obtained from sources believed to be reliable. Gartner disclaims all warranties as to the accuracy, completeness or adequacy of such information. The opinions expressed herein are subject to change without notice.
Although Gartner research may include a discussion of related legal issues, Gartner does not provide legal advice or services and its research should not be construed or used as such. Gartner is a public company,
and its shareholders may include firms and funds that have financial interests in entities covered in Gartner research. Gartner’s Board of Directors may include senior managers of these firms or funds. Gartner research
is produced independently by its research organization without input or influence from these firms, funds or their managers. For further information on the independence and integrity of Gartner research, see “Guiding
Principles on Independence and Objectivity” on its website, http://www.gartner.com/technology/about/ombudsman/omb_guide2.jsp.
CUSTOMER SPOTLIGHT:
The Results: The City of Davenport saw 6.5-to-1 data reduction as well as linked clones replication from 45 minutes down to two. Average write latency went from 20ms to 1ms, and read from 17ms to 0.5ms.
The Business Benefit: Because they decided to use virtual desktops instead of buying physical desktops, the FlashArray has already paid for itself – within their budget. And there’s still extra capacity to take on additional workloads.
“Anyone who is going to do a VMware View deployment should put their linked clones on a Pure Storage array. It’s essential.” Cory Smith, Network Administrator
CUSTOMER SPOTLIGHT:
The Results: After deploying the Pure Storage FlashArray, ARcare has achieved 5-to-1 data reduction and has reduced their data footprint by 79%.
The Business Benefit: ARcare experienced immediate performance improvements across all applications after migrating to the Pure Storage FlashArray. Latency issues were eliminated with sub-millisecond response rates for virtual desktop users despite any additional network or application activity, including patching.
“I can’t stress the FlashArray at all; it can handle anything I throw at it.” Adam Ausburn, Network & Virtualization Administrator
3
3
most vendors retrofitted 5-10% flash storage
into their architectures to attempt to speed-
up the disk arrays via caching & tiering
technologies.
While flash caching/tiering provided
some noticeable improvement, the hybrid
flash/disk array model also suffered
from increased complexity and inherently
variable performance: users simply didn’t
know whether a given storage IO would be
serviced at flash speed (<1 millisecond) or
disk speed (5-10 milliseconds). So ultimately
the storage industry realized that to make it to
the next plateau of performance, you simply
had to go “all flash,” and you really needed to
design an array from the ground-up for flash
memory, hence the birth of all-flash arrays.
In parallel, a similar evolution was happening
in the flash world. Flash entered enterprise
computing in the form of PCIe flash cards
and flash SSDs, both of which were good for
accelerating a single application in a single
server. While this was helpful for some point
Flash storage and disk storage have been
on different evolution paths, but they have
converged to create the ideal storage
platform for enterprise computing.
Disk storage has been the dominant storage
platform for the enterprise for the better
part of 50 years, but spinning disk has run
into a fundamental problem: it gets bigger
every year on a capacity basis, but it doesn’t
get any faster (limited by the fundamental
rotational physics), so it actually gets slower
every year on a performance-per-capacity
basis. As a result, the storage industry has
been desperately enhancing disk-based
storage arrays for the past decade to try
and keep up with the continuous compute
advances driven by Moore’s Law. In the
early 2000s the dominant strategy was
“more disk” – or wide striping, where more
and more disk spindles were thrown inside
arrays to improve performance. As this
became clearly inefficient and unworkable,
flash storage entered the scene as a method
for adding performance to disk arrays, so
All-Flash Arrays: The Evolution of Storage & Flash
applications, enterprises are fundamentally
built on consolidated, clustered, and resilient
application architectures leveraging shared
storage, and the server-by-server approach
to early flash had limited appeal. Flash
devices then grew into “Flash Appliances”
– slightly larger (1-20TB) flash devices that
could be networked and had additional
resiliency, enabling them to be shared by
small compute clusters. But these flash
appliances lacked the rich software layer
that was common in enterprise disk arrays:
complete resilience through RAID and
HA, snapshots, replication, encryption,
data reduction technologies, and deep
application integration. They were also
very expensive – often costing 5X the cost
of comparable enterprise disk arrays. And
so the next natural step in the evolution was
clear: marry the “all flash” performance,
space, and efficiency advantages of the
flash appliances with the enterprise-quality
features of the disk arrays to build a product
from the ground up that combines the best
of both worlds: the All-Flash Array.
4
The results are stunning: All-Flash Arrays
represent an order-of-magnitude gain in the
key dimensions that are used to evaluate
storage – performance, size/rack space,
power consumption, and management
simplicity. They help storage keep pace with
compute by moving from 10s of thousands of
IOPS of performance to 100s of thousands,
and reduce latency from 5-10ms to <1 ms.
They take the advantages of early flash
products, but deliver them in an array form-
factor that achieves the resiliency, scale, and
simplicity that allows enterprises to easily
adopt them in their existing infrastructure.
And the final, and perhaps most important
break-through has been price. By marrying
advances in consumer-grade MLC flash
memory with inline data reduction
technologies like deduplication and
compression, all-flash array vendors like
Pure Storage have been able to break the
price barrier, delivering all-flash arrays for
the same cost or even lower than legacy disk
architectures on a $/usable GB basis.
In the end the big winners in this transition
are enterprise storage customers and end
users. After living with incremental advances
in disk storage for decades that saw
compute improvements fundamentally out-
strip storage to the point where storage is the
bottleneck in most environments, customers
can now unlock a 10X improvement in the
storage architecture at constant or even
reduced spend without having to go through
a massive upheaval in their infrastructures.
Going through this transition will create
massive opportunities for improving their
operational efficiency, and accomplishing
things in business that simply weren’t
possible before.
Source: Pure Storage
CUSTOMER SPOTLIGHT:
The Results: TripPak attained 6-to-1 data reduction of servers, Citrix XenDesktops, Oracle, and SQL Databases running on vSphere 5.0.
The Business Benefit: With the FlashArray in place, TripPak customers are experiencing lightning-fast response times from their high workload database queries. TripPak IT is using less power, less space, and increasing performance by 10X, while their environment has plenty of headroom to grow as they move more workloads over to the FlashArray.
“The FlashArray is a storage guy’s dream. We love the design!” David Abbott, Sr. Infrastructure Engineer
CUSTOMER SPOTLIGHT:
The Results: Siemens saw 6.7-to-1 data reduction after implementing the FlashArray. They also saw 4X improvement per instance in VM provisioning time.
The Business Benefit: Prior to migrating to the Pure Storage FlashArray, Siemens was having issues with scalability. Now, as they grow, they can easily – and non-disruptively – scale their storage to meet the demands of their diverse set of engineering, QA, support, training, and customer demo needs.
“Even with 50 VMs running all day on one LUN, we’re not even pushing the storage.” Bryan Bond, Sr. Systems Administrator
5
Transitioning to the All-Flash Enterprise
Now that All-Flash Arrays have emerged as the evolution of storage,
most enterprises will move to a model where their core business
runs on flash within 5 years. For some of the most aggressive
this will mean that they will literally eliminate disk, as the leading
enterprises, consumer web companies are increasingly doing. For
others, that will mean that disk will still be used, but the large majority
of applications that power the business will run on an all-flash
architecture. In the end whether you actually remove every spinning
disk is beside the point: flash will power the apps that drive your
business.
In addition to the advent of the All-Flash Array, the other force
powering this change is the progression down the virtualization
journey that enterprises are taking towards private cloud
architectures. As organizations moved to virtualize their servers,
a common model was to implement “tiers of storage” – multiple
combinations of slow disk, fast disk, tape, and sometimes even flash
to deliver different service levels and storage costs for applications.
This model worked fine for early virtualization, where consolidation
was limited and where most deployments were application islands.
As virtualization pushes towards the true private cloud, deeper
infrastructure consolidation starts to happen – where large compute
farms share resources with 100s or 1,000s of heterogeneous
applications. Enabling this model requires a higher-performance
storage tier capable of mixing workloads and delivering predictable
service levels for all apps, but also fits an economic profile that makes
widespread deployment possible.
As such, we see the storage architecture of tomorrow being highly
streamlined, simplified, and consolidated into two main tiers: a
“performance tier” for business applications served by All-Flash
Arrays, and a “capacity tier” served by SATA disk, serving 2nd/3rd tier
applications, content farms, backups, and archiving. Virtualization
provides the perfect abstraction layer to enable movement of
workloads between these tiers based upon changing needs.
Source: Pure Storage
6
The technical reasons for moving to all-flash are clear, but there
has been one final (and perhaps more important) inhibitor to flash
memory adoption in the enterprise: simple economics. With flash
media historically 5-10X more expensive than disk, most enterprises
just couldn’t afford to move to all-flash broadly.
This is where some of the innovations that Pure Storage has
pioneered in the all-flash array space shine: put simply we make
all-flash storage affordable for everyone, and we make it affordable
enough to use broadly across your enterprise. By combining low-
cost consumer grade flash memory (which we make enterprise
resilient through our FlashArray architecture and Purity software) with
inline data reduction technologies (deduplication, compression, thin
provisioning), we are able to reduce the cost of 100% flash memory
to under the cost of spinning disk on a straight CapEx basis. Then, if
you are willing to consider the OpEx savings (maintenance, power,
cooling, data center rack space, and management cost), we
are able to reduce the cost even further to the point where most
deployments realize substantial savings, as shown below.
While these results are impressive and justify the all-flash storage
purchase on their own, for most customers the real advantage
comes in the up-stream impact to their business. What value do
you put on getting a competitive advantage in the marketplace?
On delivering better service for your customers? On generating
higher customer loyalty? On happier end users? On being able to
analyze substantially more data to make better-informed business
decisions? Downstream costs will help you justify the move to
flash, but the up-stream business benefits will be the real reward.
Source: Pure Storage
Solving the Flash for All Economics
7
7
The early days of flash deployments were
largely centered on accelerating point
database applications where storage IO
was the key bottleneck. Now that Pure
Storage has broken the price barrier for
flash, all-flash storage is suitable for a wide
range of enterprise applications.
Across the Pure Storage user base, we see
three main use cases where customers are
adopting flash. Because data types vary in
deduplication potential and compressibility,
data reduction rates vary slightly across
these use cases, ranging from 2-4X for
databases which are compression-driven,
up to greater than 10X for VDI which
achieves very high deduplication results.
Server Virtualization is a key driver of flash
adoption. As workloads consolidate and
virtualization ratios top 50+ VMs/server, the
IO stream down to storage becomes highly
random. Traditional disk architectures just
don’t handle random IO well, making disk
the key bottleneck in expanding server
virtualization. To make matters worse,
virtualization deployments often start with
the “easiest” VMs, so as customers expand
to virtualized Tier 1 applications, the IO
bottleneck gets worse. Flash is the answer
to keep on the path to enterprise-wide
virtualization.
Databases and Analytics are some of
the most obvious fits for flash. Storage
IO is often the key wait time in enabling
databases to process more transactions,
or analytic queries to complete faster. In
addition to processing more transactions
and analyzing more data at lower storage
cost, moving to flash storage often enables
massive consolidation at the application tier.
Since each database instance can now do
more work, enterprises can typically reduce
instances, leading to direct cost savings
in terms of both hardware and database
license savings.
Virtual Desktop Infrastructure is perhaps
the perfect use case for flash. It requires
a very high level of performance as disk
IO is directly in the end-user experience
path, it drives highly random IO due to the
high consolidation, and cost is at a very
high premium, since VDI ROI cases are
fragile at best. Leveraging all-flash storage
provides the absolute best end-user desktop
experience, and since VDI is one of the most
reducible workloads in the data center, the
FlashArray is able to drive the cost of all-flash
VDI to well under the cost of spinning disk.
Source: Pure Storage
Flash for All Applications
8
The Business Benefits of Flash for All
Achieving Business Results that Are Not Even Imaginable with Disk
We have to build our businesses and
compute infrastructures around the tools
we have in a given generation, and the
capabilities and limitations of these tools
define both what we achieve, but on a
more fundamental level, our expectation of
what is possible. Moving to flash will have
an immediate, quantifiable improvement
in your existing business infrastructure:
process more transactions, analyze more
data, answer questions faster: and be
more agile and responsive. But a whole lot
more is possible: results that weren’t even
imaginable when you limited your thinking
to disk.
What that hidden advantage is and
how it turns into revenue will vary widely
depending on your unique business and
industry, but imagine how you’d fare on the
playing field vs. your competition with an all-
flash architecture powering your business.
Imagine taking the storage bottleneck out
of the compute chain, accelerating each of
your core business applications dramatically.
Imagine the levels of analysis possible,
accessing larger data sets with more
complex queries, asking questions and
finding answers in your data that were stuck
inaccessible on disk.
Now imagine if the situation was reversed,
and you choose not to take the path of flash
when one of your key competitors does.
How will it change the dynamics of your
market, and which side of the playing field
do you want to occupy?
Deliver Differentiated, Compelling End-User Experiences for All
Most computing today is done in the
service of some ultimate end-user, typically
either an employee or a customer of your
company. While moving to flash has the
tangible business benefits we discussed
earlier, it also has a dramatic impact on the
experience of users, and happier employees
and end-users are directly connected
to profitable businesses. Now flash for
all enables you to deliver the all-flash
experience to all the end users you serve.
On the employee side, flash for all is about
unlocking the productivity and creativity of
your workforce, and delivering a computing
experience that results in delight vs.
frustration. This has hard-cost elements (less
calls to the help desk, less working around
IT), but the real benefits come in better
empowerment and employee engagement.
For most business, employees are the core,
and the relative cost of empowering your
employees with the absolute best computing
experiences possible pales in comparison
to both what you pay your employees, and
the productivity gain you’ll achieve from
these investments. Delivering anything less
than the all-flash user experience for your
employees just doesn’t make economic or
business sense.
And on the end-user customer side, study
after study has shown a direct link between
customer satisfaction and loyalty to business
profitability and growth. Flash for all
enables you to create new and differentiated
customer experiences. Think: search results
that appear before you finish typing, a world
of information accessible in the palm of
your hand, immersive gaming experiences,
and customer support that is always a step
ahead. How can you win the hearts, minds,
and loyalty of your customers with flash?
Achieve New Levels of Efficiency in the Storage Factory
If you dive into your organization and talk
to folks high and low about storage, you’ll
probably find a disturbing reality:
Your Storage Admins don’t really like
their current storage array. Why is that?
As storage has failed to keep up with the
compute infrastructure, Storage Admins are
often the people stuck in the middle between
increasing business demands, flat budgets,
calls about slow performance. Disk has
simply gotten way too complex over the years,
it is failing to meet the business demands,
and they are left to take the blame and solve
the problem, usually with more disk.
Your CIO doesn’t really like disk storage.
When the CIO looks at the current disk storage
platform and how it is contributing to his/
her objectives (adopting the private cloud,
greening the datacenter, achieving results for
the business), again disk finds itself on the
“red” side of the equation more often than
not. Disk is preventing server virtualization,
sapping too much power in the datacenter,
and holding back the business.
Flash provides an answer to both these
challenges, it is the building block of a
fundamentally more efficient storage
architecture: one that turns Storage Admins
into heroes and gives them time and sanity
back in their days, and one that helps CIOs
meet their annual objectives, helping IT
contribute to the business as opposed to
being a cost center. A transition to flash
has the opportunity to change the storage
equation from one that is a 360° liability to a
technology that enables.
Source: Pure Storage
9
Flash for All: It’s Time
Why Buy Disk?
If you are involved with the process of running, buying, or being an
end user of enterprise storage, we hope this short paper has given
you a window into the opportunity that Flash for All presents to your
business, and maybe you’ll think twice before blindly signing another
storage contract. You have the opportunity to buy storage that can
contribute to the efficiency of your organization, enable a competitive
advantage for your business, and give you the technology to deliver a
compelling all-flash experience for your users. We hope you consider
Pure Storage for your next storage purchase, but most of all we hope
you make the choice to try flash. Once you’ve experienced flash, you’ll
likely never go back.
We’re So Confident in Flash for All, It’s Guaranteed
We’re so confident in the Pure Storage FlashArray that all FlashArray
purchases come with the storage industry’s broadest guarantee:
the Love Your Storage Guarantee. If our storage doesn’t meet our
high promises and your high expectations, you can return it for a
full refund. Read more about the guarantee at purestorage.com/
guarantee.
Source: Pure Storage
CUSTOMER SPOTLIGHT:
The Results: By switching their storage to the Pure Storage FlashArray, Riverview Hospital was able to achieve 8-to-1 data reduction with hundreds of persistent desktops. They increased virtualized SQL database IOPs by 2X, and reduced latency from 2-8ms down to 0.5ms.
The Business Benefit: Through the deployment of end-to-end flash-based storage, Riverview improved the productivity of its caregivers by minimizing delays in accessing patient data.
“Our most important clinical systems now run on Pure.” Jason Pearce, Enterprise Architect
CUSTOMER SPOTLIGHT:
The Results: Yodle achieved a 5-to-1 data reduction of PostgreSQL DB despite entirely new DB at every restore. Through the Pure Storage FlashArray deployment, they were able to virtualize their database environment and add analytics processing to the same environment.
The Business Benefit: By replacing their server and adding a Pure Storage FlashArray, Yodle dropped their rebuild process from 24 to 3 hours for the development environment, and from 7 to one for the analytics rebuild – huge time-savings for their team.
“The biggest surprise was what a dramatic different we saw in build time.” John Merryman, CTO
10
Cool Vendors in Storage Technologies, 2013
From the Gartner Files:
This research details five companies that
provide innovative storage capabilities and/
or contain the costs associated with storage
and its operational management. Gartner
also takes a look back at two vendors that
were profiled in a previous Cool Vendors
report.
Key Findings
• CTERA Networks offers unique, unified
cloud storage integration hardware
and software (including EFSS) within an
expanding partner ecosystem.
• Delphix can significantly help contain
the storage requirements and overall
cost and time associated with reducing
the typical five to 12 full copies of each
production database.
• Panzura addresses the difficult global
file-locking problem with its distributed
file system appliances supporting cloud
storage.
• Pure Storage enhances inexpensive
consumer PC SSDs through flash-
optimized, in-line data reduction and
management software, delivering a
simple, cost-effective, all-flash alternative
to HDD/SSD storage arrays.
• Scality offers software based scale-out
storage with heterogeneous API support
to tackle a variety of big data and cloud
workloads.
Recommendations
• Include CTERA Networks when
implementing an integrated private or
public cloud storage environment to
support branch offices and mobile users.
• Examine Delphix’s solution to deliver
improved database service levels and
capabilities, and as part of a prudent cost
containment and efficiency strategy.
• Evaluate Panzura to implement scalable,
secure global file access and storage
consolidation among multiple sites and to
back up and archive to the cloud.
• Consider Pure Storage when application
workloads demand a cost-effective,
all-flash system solution and can benefit
from in-line data reduction.
• Consider Scality when building scale-out
object storage architectures for cloud
storage and active archiving.
Analysis
This research does not constitute an
exhaustive list of vendors in any given
technology area, but rather is designed to
highlight interesting, new and innovative
vendors, products and services. Gartner
disclaims all warranties, express or implied,
with respect to this research, including any
warranties of merchantability or fitness for a
particular purpose.
What You Need to Know
IT managers often share common goals,
including a need to modernize their storage
infrastructures and improve quality of service
while simultaneously containing costs.
New storage components, solutions and
management tools can help stakeholders
build easier-to-manage, more efficient and
more available storage infrastructures. Many
organizations are evaluating technologies
that will drive efficiency, such as higher-
performing and highly automated storage,
cloud-based solutions, and/or management
products that allow for proactive problem
avoidance and increased resource utilization
and optimization.
This research details five emerging private
vendors that can assist organizations in
meeting their IT modernization and cost
containment initiatives.
CTERA Networks
Palo Alto, CA (www.ctera.com)
Analysis by Gene Ruth
Why Cool: CTERA Networks is a privately
held company that offers cloud storage
gateway and related infrastructure products
to connect datacenters and branch offices
to public or private storage clouds. The
offerings are suitable for small or midsize
businesses (SMB) and enterprise customers
as well as service providers (telcos,
management service providers [MSPs])
and integrators in enabling cloud storage
solutions.
The CTERA Networks offerings address
file sync and share, backup and primary
data delivery using on-premises hardware
optimized to transport data to a public or
private cloud environment. The company
uniquely offers a cost-efficient solution
to support remote and branch office
(ROBO) environments on a large scale.
The lightweight local footprint of a CTERA
Networks storage unit, the elimination
of additional backup software and the
resultant operational simplification make
CTERA Networks an attractive solution for
supporting extensive ROBO environments.
CTERA Networks is positioning itself to
address the quickly expanding cloud storage
integration market with a low-cost solution
and significant array of partnerships. The
partnership ecosystem continues to expand
with native integration of CTERA Networks’
capability into the CISCO Integrated Service
Routers Generation 2 (ISR), integration with
EMC’s Atmos, IBM’s General Parallel File
System (GPFS), Hitachi Data Systems’ Hitachi
Content Platform (HCP), Dell DX, OpenStack
Swift and many relationships with telcos and
MSPs in Europe and the U.S. The addition of
a file sync and share capability enhances
the value proposition and provides a unique
competitive advantage, which dovetails
nicely with the cloud storage paradigm and
11
11
file collaboration opportunities. Additionally,
CTERA Networks also provides cloud storage
orchestration and automation software to
enable service providers providers to deliver
multitenant public cloud storage services
similar to Amazon S3.
Challenges: CTERA Networks is a relatively
new company, founded in 2008 and
competing in an intensely active segment of
the market. Their presence is growing in the
U.S. with SMB customers (from a revenue
perspective) and in Europe with a number
of large telco and communication service
provider (CSP) partners. As a small but fast-
growing company, risk exists that CTERA
Networks may struggle to adequately deliver
in a timely manner to its diverse customer
set. In addition, the cloud storage and
enterprise file synchronization and sharing
(EFSS) markets are nascent and, thus,
customers should be cautious and expect
that many of the participating vendors will
be acquired or fail, with only a few vendors
remaining independent over the next five
years.
Who Should Care: CTERA Networks should
be of interest to any IT organization that is
looking for a cost-effective unified solution
to support a public or private cloud storage
environment with file sync and share
capability behind their own firewall or as a
branded service provider to support ROBO
environments and mobile users.
Delphix
Menlo Park, CA (www.delphix.com)
Analysis by Dave Russell
Why Cool: Delphix’s Agile Data Platform
is a software product that can significantly
speed up operations and the amount of
time it takes to test production databases,
run reporting and analytics and refresh data
from days to minutes, all while applying
compression and even avoiding the need
for deduplication (since no redundant blocks
are ever copied) to shrink the footprint
of required storage. Delphix’s database
virtualization solution can dramatically
reduce the number of physical copies of a
single database, thus significantly improving
capacity utilization by decreasing the
amount of storage and power consumed.
Perhaps of even greater interest is the
very fast availability (restore) and test and
development support that the solution can
provide.
Currently, Delphix supports Oracle
Databases (including those running
on Exadata) and Microsoft SQL Server
databases, with plans to add Postgres and
Sybase in 2013 and MySQL in 2014. The Agile
Data Platform makes a one-time full copy
of the application’s blocks of data onto any
storage device, applying compression that
typically yields a 2.7x space reduction. Then,
periodic log shipping and block change
tracking keeps the Delphix copy updated.
Pointer redirection allows for any point in
time (APIT) recovery and can utilize a single
copy to rapidly serve up space-efficient full
instances of database images for testing
or recovery in minutes versus hours. The
product installs as a virtual appliance
without any agents, sits out of band and
provides IP-based replication to another
Delphix instance if desired.
Delphix has been in business since 2008,
with over $45 million in funding. It has been
shipping product since 2011, with over 100
customers including Comcast, Facebook,
Proctor and Gamble and Informatica. The
company claims over 100 employees,
with engineers that previously worked on
the Oracle Sun ZFS filesystem, Oracle’s
Recovery Manager (RMAN) and Data Guard,
VMware’s replication solution and EMC’s
Avamar. Delphix has partnerships with
SAP and system integrator Deloitte, which
includes SAP selling the solution through its
5,000-plus person sales force.
Challenges: Any emerging vendor typically
raises some degree of concern for an IT
organization, and this will likely be very
much the case for a provider that typically
targets very large enterprises and has an
average transaction price of over $250,000.
Organizations will want to perform due
diligence to get comfortable with a solution
that connects with production applications.
Other challenges are that database
administrators will need to be comfortable
with the product having the credentials to
make periodic calls to the database and
the future concept of database vendor(s)
potentially adding virtualization and data
reduction capability natively in its product.
Who Should Care: While Delphix
predominantly targets the CIO or vice
president of infrastructure and application,
Gartner believes that directors of storage
and storage administrators that are
responsible for providing database capacity
and availability should examine Delphix’s
solution as a means of improving service
levels through reduced provisioning times
and as part of a prudent cost containment
and efficiency strategy.
Panzura
Campbell, CA (www.panzura.com)
Analysis by Pushan Rinnen
Why Cool: Founded in July 2008 in San
Jose, CA, with a total venture funding of $33
million, Panzura offers global distributed file
system appliances with multisite global file
locking and global deduplication, local file
pinning, and cloud storage integration. The
company has gained market traction since
product launch with over 100 customers by
the end of 2012, 50% of which are multi-
billion-dollar organizations.
The Panzura architecture is primarily
designed to build a private network
in conjunction with public clouds. The
Quicksilver Cloud Storage Controllers are
12
a series of virtual and physical network-
attached storage (NAS) appliances that have
solid-state drives (SSDs) and hard drives
for local cache (up to 36TB raw today) to
store the most frequently accessed data,
while pushing the majority of infrequently
accessed data to the private or public cloud.
The file-locking mechanism embedded in
its global namespace solves the problem
of global cache coherency among multiple
sites and can scale better than Microsoft’s
DFS. Panzura’s primary use case is global
content access and cross-geo project
sharing by consolidating NAS devices and
local file servers. Because all sites see the
same file system and site controllers can be
added to the same network transparently,
the solution can speed up global workflow
process and make the distributed storage
infrastructure more efficient. The Panzura
solution is also used by customers to back
up and archive data to the cloud, because
Panzura supports high-throughput capability
for Symantec’s NetBackup and IBM’s Tivoli
Storage Manager, as well as archiving
tools such as Atempo Digital Archive and
Symantec Enterprise Vault.
The Panzura solution supports most major
cloud storage providers at the back end
such as Amazon, Nirvanix, Google, as well
as EMC’s Atmos-based cloud providers and
public clouds from Dell, HP and IBM. Most
of these partners also resell the Panzura
solution, which offers AES-256 encryption
for data at rest and TLS for data in flight
and, with its FIPS 140-2 government security
certification, has gained several large
federal government customers. Panzura’s
other customers include companies from the
media/entertainment, healthcare, financial
services and research and education fields.
Challenges: Panzura faces the challenge
of debunking the common perception that
cloud storage is only for small businesses
and small workloads. It also needs to
expand its ecosystem by adding more
support for on-premises applications.
When Panzura is used to support traditional
applications such as backup, it may have
compatibility issues, resulting in certain
application functions not working properly.
On the global file access front, Panzura lacks
the mobile device support, which is offered
by file sync and share services. The company
needs to expand its ecosystem by adding
more independent software vendor (ISV)
support in the backup and archiving arena.
Who Should Care: Panzura should appeal
to midsize to large organizations that want
to implement private or hybrid cloud storage
to consolidate multisite file storage, offer
more effective project and data sharing,
and provide backup and archiving to the
inexpensive public cloud.
Pure Storage
Mountain View, CA (www.purestorage.com)
Analysis by Joseph Unsworth and John
Monroe
Why Cool: Pure Storage was founded in
2009 on the premise of building an in-
line, flash-optimized deduplication and
compression software architecture around
relatively inexpensive PC SSDs to deliver a
cost-effective SSD appliance. Pure Storage
leverages the commodity multilevel cell
(MLC) SSD form factor in its arrays, today
sourcing SSDs from leading PC SSD supplier
Samsung; however, Pure’s innovation is
centered around their core competency in
software.
Pure Storage has over 30 patents from areas
such as deduplication, flash management
and resiliency, which are all critical elements
when solely relying on flash memory
technology for persistent storage in data
center environments. Pure Storage boasts
flash-specific software features with the
most interesting being in-line deduplication/
compression typically at a rate of 5-to-
1, high-availability with nondisruptive
upgrades, snapshots and replication that
represent minimal overhead on system
performance and latency. With system
price-points ranging around $10/GB for
usable capacity and Pure Storage’s strategy
of enhancing the most capable, cost-
effective PC grade SSD, they have distanced
themselves from the challenges of flash
hardware management, opting instead to
rely on their core competencies in software.
Pure Storage has raised $95 million in
aggregate, including its latest series-D
funding in 2012. This funding ensures
financial stability in the near-term and
the ability to scale in a dynamic, ultra-
competitive market that is still in its infancy.
Pure Storage has over 200 customer wins,
a testament to not only their software but
the simplicity with which it can be integrated
within existing storage architectures.
Challenges: Pure Storage’s greatest
challenge will be articulating its value
proposition clearly and effectively as a
startup in a landscape fraught with hype
and confusion disseminated by a crowd of
new and traditional storage array vendors.
Pure Storage does not have the fastest nor
largest capacity system (up to 40TB per
system) in the market, but it is the software
differentiation and optimization selling
points that will need to be reinforced along
with customer wins. This will be particularly
important since Pure Storage is relying only
on value-added resellers (VARs) and reseller
partnerships as a path to market.
Who Should Care: Data center professionals
who demand higher storage platform
performance supported by capable
data management software features.
The application workload demand and
complexity will dictate the need for such
solutions, especially in highly virtualized
environments such as virtual desktop
infrastructures (VDIs), front-end Web
indexing, online transaction processing
(OLTP) databases, data warehouses and
high-performance computing applications.
13
13
Scality
San Francisco, CA (www.scality.com)
Analysis by Arun Chandrasekaran and Alan
Dayley
Why Cool: Scality is a venture-backed,
privately held company headquartered
in Silicon Valley and an R&D team in
France that offers scale-out, software-
based storage that can potentially deliver
massive scalability, high availability and
data protection at lower costs. While
Scality’s primary focus has been on the
object storage space, the vendor has an
aggressive road map for unified storage
with block and scale-out file system
capabilities targeting enterprise workloads
that need high performance.
Scality’’s RING Organic Storage topology
is based on a peer-to-peer architecture,
which means that there are no master
nodes and no single point of failure. Both
data and metadata is distributed across
the nodes in a shared-nothing model with
parallel infrastructure and operations (I/O)
paths. The architecture provides near-
linear performance aggregation, allowing
customers to achieve incremental capacity
and performance gains at a lower cost.
By packaging multiple storage nodes and
I/O daemons at a server level, and due to
its unified storage architecture, Scality is
able to potentially target more versatile and
performance-oriented workloads that are
beyond the purview of many object storage
vendors.
Scality offers simple software replication
and configurable erasure coding that can
significantly reduce storage overheads,
provide self-healing and improve data
rebuild times, which may improve costs,
availability and manageability on a PB scale.
Given the fragmented nature of APIs in the
object storage space, Scality has taken
a more neutral approach by supporting
APIs from Amazon S3, OpenStack Swift
and CDMI, apart from its proprietary API.
This heterogeneous API support will be a
critical need for customers planning hybrid
cloud storage implementations. Scality’s
integration work with Apache Hadoop and
OpenStack expands its addressable market
opportunity although it faces significant
competition from more-established vendors
in each of these initiatives.
Challenges: Scality is in direct competition
not only with other object storage vendors,
many of whom have been around for a
longer time, but also with large unified
storage vendors such as EMC and NetApp.
Hence, Scality faces huge barriers to entry
posed by the significant R&D, marketing
prowess and deep relationships that the
incumbent storage vendors possess.
Additionally, Scality has a distributed team
with R&D based in France and sales,
marketing and operations in the U.S.,
which has the potential to affect time to
market and possibly create cultural gaps.
While there has been a lot of buzz about
“software-defined storage” of late, Scality’s
software form factor is an inherently difficult
value proposition to sell in an industry
where storage is normally associated with
hardware, which needs a very different
distribution, support and sales model.
Who Should Care: IT leaders, technical
architects and application owners in
both enterprise and service provider
environments looking for scalable, self-
healing and cost-effective storage for cloud
and big data workloads should consider
Scality. Gartner expects the initial interest
to come from cloud service providers
evaluating software as a service (SaaS)
storage and storage as a service, and
enterprises looking for an active archive
solution.
Where Are They Now?
Nimble Storage
San Jose, CA (www.nimblestorage.com)
Analysis by Arun Chandrasekaran
Why Cool Then: Nimble Storage’s flagship
product is an innovative Internet Small
Computer System Interface (iSCSI) based
storage array that is optimized for flash.
Nimble’s architecture, known as Cache-
Accelerated Sequential Layout (CASL),
achieves high write performance by turning
random I/O into sequential I/O. CASL does
this by compressing data into variable-size
blocks and combining them into full stripes
that are written to disks. It then uses SSDs
to cache active data to achieve high read
performance. CASL’s variable block sizes
enable the system to be tuned for different
applications, to support various compression
algorithms and to allow each block to store
additional metadata, such as checksums to
detect bit errors. The product line known as
the CS-Series is addressing the midmarket’s
need to modernize its storage infrastructure,
and its backup and disaster recovery
process.
The CS-Series offers in-line compression
(which applies to primary data and local
and replicated snapshots), thin provisioning,
highly efficient snapshots, flexible replication
and cloning tools, as well as data protection
policies and redundant array of independent
disks (RAID) 6.
Where Are They Now?: When Gartner
profiled Nimble Storage in early 2011, it had
less than $10 million in revenue and the
product was barely six months old. Nimble
has since expanded the product line with
new entry-level and high-end models,
and significantly grown the revenue and
customer footprint. Nimble Storage now
boasts an annual bookings run rate in their
latest fiscal year of more than $100 million
and more than 1,250 customers worldwide,
14
making it one of the fastest-growing storage
startups. Employee count has increased
from less than 50 people in early 2011 to
close to 400 employees today. From starting
as a U.S.-centric company, Nimble Storage
has expanded its direct and indirect footprint
in EMEA and Asia/Pacific, with close to 15%
of annual revenue coming from outside the
U.S. Nimble Storage is planning for an initial
public offering (IPO) in late 2013 or early 2014
to raise more capital for business expansion.
Nimble has also forged close partnerships
with vendors such as Cisco, VMware,
Microsoft and CommVault to expand its
addressable use cases by tackling a variety
of workloads such as VDI, Exchange,
SharePoint and Oracle and Microsoft SQL
Server databases.
Despite this tremendous growth, Nimble
still faces significant challenges. Apart
from viability concerns that most startup
vendors encounter, Nimble’s product
line is still narrowly focused on a few use
cases and there is still work to be done in
bridging the channel and ISV partnership
gaps. Additionally, Nimble is focused only
on iSCSI protocol, when growth in NAS is
far outstripping the growth in storage area
networks (SANs).
Who Should Care: Midsize organizations
looking for flash-optimized storage for
workloads such as Oracle Database, SQL
Server database, SharePoint and VDI should
include Nimble Storage in their shortlist.
Nimble’s products offer the promise of
eliminating the complexity of managing
separate primary, backup and disaster
recovery processes and infrastructures while
delivering performance enhancements
based on a flash-optimized architecture.
StorSimple
Santa Clara, CA (www.storsimple.com)
Analysis by Arun Chandrasekaran and
Stanley Zaffos
Why Cool Then: StorSimple Cloud Storage
Controller is an IP storage gateway that
behaves as an iSCSI storage system that
uses public cloud storage as a storage tier.
Instead of destaging data from cache to
back-end storage, it destages data into the
cloud after it deduplicates and encrypts
the data. From a storage perspective, it is a
disaggregated system with everything but
back-end storage in the appliance. From a
network perspective, it is a WAN optimization
controller (WOC) supporting public or private
cloud-distributed storage.
Internally, the StorSimple Storage Controller
uses a weighted storage layout approach,
data is broken into “chunks” that are
analyzed and weighted based on frequency
of use. SSD capacity is split between a linear
store and a deduplicated store. Data in the
linear store is not deduplicated and holds
the hottest data. The deduplicated store, as
its name suggests, holds deduplicated but
unencrypted data and is used to expand
the usable capacity of the controller. When
working, set management algorithms
determine that data should be destaged to
the cloud, it is encrypted and then moved
from SSD and SAS layers to low-cost cloud
storage infrastructure-as-a-service (IaaS)
providers such as Amazon Web Services
(AWS) and Microsoft Windows Azure.
StorSimple provides block-level snapshots,
enabling recovery point objectives and
backups not generally associated with
a cloud IaaS infrastructure. Although
there were numerous startups that were
developing similar on-site/cloud hybrid
storage approaches, StorSimple was the
first developer to optimize for specific
applications with iSCSI access.
Where Are They Now?: Microsoft acquired
StorSimple for an undisclosed sum in
November 2012, validating StorSimple’s
innovative cloud-integrated storage
strategy. Microsoft’s underlying intention is
to make hybrid cloud storage a reality for
its enterprise and midmarket customers
through a combination of physical and virtual
on-premises appliances. The acquisition
gives Microsoft a significant advantage over
competitors such as Amazon Web Services,
AT&T and Rackspace in targeting Microsoft
workloads such as SharePoint and SQL
Server.
Since the acquisition, Microsoft has doubled
the engineering team at StorSimple and
significantly expanded its global reach
beyond North America, with the product now
being available in more than 25 countries.
The acquisition has also helped StorSimple
to rapidly expand its customer base due to
tight integration and joint go-to-market with
Windows Azure for enterprise workloads
such as SharePoint.
Although post acquisition, Microsoft had
indicated that it had no plan to change
the existing integration with other cloud
providers such as Amazon and Rackspace,
we recommend that customers should get
clear direction from Microsoft on its long-
term support plans and feature availability
for non-Microsoft clouds.
Who Should Care: VPs and directors of
IT, storage, and infrastructure of midsize
companies and enterprises planning a cost-
effective hybrid cloud storage environment
with iSCSI access for primary storage,
backup or archiving use cases should
consider StorSimple.
Source: Gartner Research, G00250240,
Dave Russell, Arun Chandrasekaran, Gene Ruth, Joseph
Unsworth, John Monroe, Pushan Rinnen, Alan Dayley,
Stanley Zaffos, 22 April 2013
15
15
Pure Storage, Inc.
Twitter: @purestorage
650 Castro Street, Suite #400
Mountain View, CA 94041
T: 650-290-6088
F: 650-625-9667
Sales: [email protected]
Support: [email protected]
Media: [email protected]
General: [email protected]
About Pure Storage