Cloud Bursting For Dummies Guide

CloudBursting
by Sharif Nijim
Avere Systems Special Edition

Cloud Bursting For Dummies®
, Avere Systems Special Edition
Published by
John Wiley & Sons, Inc.
111 River St.
Hoboken, NJ 07030‐5774
www.wiley.com
Copyright © 2015 by John Wiley & Sons, Inc., Hoboken, New Jersey
No part of this publication may be reproduced, stored in a retrieval system or transmitted in any
form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise,
except as permitted under Sections 107 or 108 of the 1976 United States Copyright Act, without the
prior written permission of the Publisher. Requests to the Publisher for permission should be
addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ
07030, (201) 748‐6011, fax (201) 748‐6008, or online at http://www.wiley.com/go/permissions.
Trademarks: Wiley, For Dummies, the Dummies Man logo, The Dummies Way, Dummies.com, Making
Everything Easier, and related trade dress are trademarks or registered trademarks of John Wiley &
Sons, Inc. and/or its affiliates in the United States and other countries, and may not be used without
written permission. Avere, FlashCloud, FlashMirror, and FlashMove are trademarks or registered
trademarks of Avere Systems Inc. All other trademarks are the property of their respective owners.
John Wiley & Sons, Inc., is not associated with any product or vendor mentioned in this book.
LIMIT OF LIABILITY/DISCLAIMER OF WARRANTY: THE PUBLISHER AND THE AUTHOR MAKE
NO REPRESENTATIONS OR WARRANTIES WITH RESPECT TO THE ACCURACY OR
COMPLETENESS OF THE CONTENTS OF THIS WORK AND SPECIFICALLY DISCLAIM ALL
WARRANTIES, INCLUDING WITHOUT LIMITATION WARRANTIES OF FITNESS FOR A PARTICULAR
PURPOSE. NO WARRANTY MAY BE CREATED OR EXTENDED BY SALES OR PROMOTIONAL
MATERIALS. THE ADVICE AND STRATEGIES CONTAINED HEREIN MAY NOT BE SUITABLE FOR
EVERY SITUATION. THIS WORK IS SOLD WITH THE UNDERSTANDING THAT THE PUBLISHER IS
NOT ENGAGED IN RENDERING LEGAL, ACCOUNTING, OR OTHER PROFESSIONAL SERVICES. IF
PROFESSIONAL ASSISTANCE IS REQUIRED, THE SERVICES OF A COMPETENT PROFESSIONAL
PERSON SHOULD BE SOUGHT. NEITHER THE PUBLISHER NOR THE AUTHOR SHALL BE LIABLE
FOR DAMAGES ARISING HEREFROM. THE FACT THAT AN ORGANIZATION OR WEBSITE IS
REFERRED TO IN THIS WORK AS A CITATION AND/OR A POTENTIAL SOURCE OF FURTHER
INFORMATION DOES NOT MEAN THAT THE AUTHOR OR THE PUBLISHER ENDORSES THE
INFORMATION THE ORGANIZATION OR WEBSITE MAY PROVIDE OR RECOMMENDATIONS IT
MAY MAKE. FURTHER, READERS SHOULD BE AWARE THAT INTERNET WEBSITES LISTED IN
THIS WORK MAY HAVE CHANGED OR DISAPPEARED BETWEEN WHEN THIS WORK WAS
WRITTEN AND WHEN IT IS READ.
For general information on our other products and services, or how to create a custom For Dummies
book for your business or organization, please contact our Business Development Department in
the U.S. at 877‐409‐4177, contact info@dummies.biz, or visit www.wiley.com/go/custompub.
For information about licensing the For Dummies brand for products or services, contact
BrandedRights&Licenses@Wiley.com.
ISBN 978‐1‐119‐21254‐6 (pbk); ISBN 978‐1‐119‐21255‐3 (ebk)
Manufactured in the United States of America
10 9 8 7 6 5 4 3 2 1
Publisher’s Acknowledgments
We’re proud of this book and of the people who worked on it. For details on how to create a custom
For Dummies book for your business or organization, contact info@dummies.biz or visit www.
wiley.com/go/custompub. For details on licensing the For Dummies brand for products or
services, contact BrandedRights&Licenses@Wiley.com.
Some of the people who helped bring this book to market include the following:
Project Editor: Martin V. Minner
Acquisitions Editor: Amy Fandrei
Editorial Manager: Rev Mengle
Avere Systems Contributors:
Joy Burd and Jeff Tabor
Business Development Representative:
Sue Blessing
Production Editor: Kinson Raja

Table of Contents
Introduction........................................................ 1
About This Book......................................................................... 1
Icons Used in This Book............................................................. 1
Beyond the Book......................................................................... 2
Where to Go from Here.............................................................. 2
Chapter 1: Understanding NAS. . . . . . . . . . . . . . . . . . . . . . 3
Getting to the Basics of NAS Storage........................................ 3
Handling the Incredible Growth and
Management of Data............................................................... 5
Finding an Efficient Solution to Overloaded
NAS Systems............................................................................ 6
Chapter 2: Understanding Cloud. . . . . . . . . . . . . . . . . . . . . 7
Appreciating the Cloud’s Advantages...................................... 8
Ah, Yes — the Cloud’s Disadvantages..................................... 8
Learning the Lingo: Cloud Terminology................................ 10
Namespace...................................................................... 10
Buckets............................................................................. 11
Objects............................................................................. 11
What’s in a Name?..................................................................... 12
Parts of the Cloud..................................................................... 13
Compute servers............................................................. 14
Storage............................................................................. 14
Network............................................................................ 15
Looking at Cloud Types........................................................... 15
Private clouds................................................................. 15
Public clouds................................................................... 15
Hybrid clouds.................................................................. 16
Accessing and Securing the Cloud......................................... 16
Transitioning into the Cloud................................................... 17
Chapter 3: Introducing Avere Hybrid Cloud NAS. . . . . . 19
Bringing NAS and the Cloud Together As One...................... 19
Using the Best of Both NAS and the Cloud............................ 23
Making things faster....................................................... 23
Making things cheaper................................................... 24
Making things simpler.................................................... 25

iv Cloud Bursting For Dummies, Avere Systems Special Edition
These materials are © 2015 John Wiley Sons, Inc. Any dissemination, distribution, or unauthorized use is strictly prohibited.
Easy Data Migration.................................................................. 26
Protecting Your Data................................................................ 27
Connecting to the Cloud.......................................................... 28
Making Things Better............................................................... 30
Chapter 4: Reaping the Benefits of
Cloud Bursting. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Benefits....................................................................................... 31
Faster time to market..................................................... 31
Pay only for what you use............................................. 32
Focus on your business, not IT..................................... 32
Improve your financial agility....................................... 33
Usage Models............................................................................ 33
Temporary usage at peak times................................... 33
Permanent usage, full time............................................ 34
Data Can Be Anywhere............................................................. 34
Speed of Flash, Cost of Disk..................................................... 34
Scale, Baby, Scale...................................................................... 35
Chapter 5: Looking at Cloud Bursting Use Cases. . . . . 37
Cloud Bursting Is Everywhere................................................. 37
Movie Rendering....................................................................... 38
Market Risk Analysis................................................................ 39
Genomic Sequencing................................................................ 40
The Sky is the Limit.................................................................. 41
Chapter 6: Ten Benefits of Hybrid Cloud NAS. . . . . . . . 43

Introduction
The World Wide Web’s influence on traditional business
computing has caused an explosive rate of data growth.
By itself, data is kind of useless — when analyzed using
computers, it can add value to your business. The ability to
seamlessly have fast access to data stored in on‐premises net-
work attached storage (NAS), private object storage, or public
cloud object storage is critical to achieving strategic goals.
Facing demand fluctuations, companies need ways to meet
that demand without expanding data centers. By leveraging
the advantages of the cloud, companies can increase storage
and compute capacity while saving money. If on‐premises
storage performance in a pay‐as‐you‐go supplemental infra-
structure world sounds appealing, this book is for you.
About This Book
Cloud Bursting For Dummies, Avere Systems Special Edition,
gets you up to speed on network attached storage (NAS),
hybrid cloud NAS, and cloud bursting. Discover how cloud
compute farms and storage technologies, combined with
on‐premises NAS infrastructure, can create a hybrid cloud
environment. As on‐premises equipment blends with what’s
available in the cloud, hybrid cloud helps you optimize your
adaptation of cloud technologies with your existing compute
and NAS storage infrastructure.
Icons Used in This Book
As you go through this book, keep a lookout for the following
icons in the margins. They denote tasty tidbits to help keep
you from daydreaming. Of course, if you are daydreaming,
that’s great — your head is already in the clouds!
This icon denotes useful bits of information to squirrel away
in the back of your mind. You’ll also find a few factoids sprin-
kled in that you can use to wow your friends.

2 Cloud Bursting For Dummies, Avere Systems Special Edition
This identifies a taste of the technical. Not enough detail to
be a hybrid cloud engineer, just enough detail to help you ask
the right questions.
This icon points out things to consider as you plan your
hybrid cloud strategy.
This icon highlights pitfalls to watch out for — history is filled
with lessons you don’t want to relearn.
Beyond the Book
Avere Hybrid Cloud NAS is reinventing the way you store and
access critical business data. Check out www.averesystems.
com for further information on hybrid cloud NAS.
Where to Go from Here
If you’ve never heard of NAS or cloud computing, it’s a good
idea to take a look at Chapters 1 and 2 first to get an intro into
NAS and the cloud. Otherwise, feel free to skim the book and
dive into whichever chapter interests you most. Beyond that,
buckle up, sit back, relax, and enjoy the read.

UnderstandingNAS
In This Chapter
▶▶ Reviewing network attached storage
▶▶ Managing data growth
▶▶ Looking for ways to resolve overloaded NAS systems
Network attached storage (NAS) has been around for a
long time. In fact, as soon as early system architects
decided to build a network between their computers so that
they could share information, they essentially built the pre‑
cursor to today’s NAS systems. After these architects sent
some bits and bytes from one computer to another, they
decided it would be a great idea to keep that data around for
a while — which led to the remote storage of data, separate
from the computer that actually created it. It also led to the
accumulation of data that didn’t need to be stored for long,
if at all. Because people like to play it safe and keep data
around “just in case,” the problem of data management
was born.
In this chapter, I review what NAS is, give you some insight
into how managing data nowadays can be difficult, and
offer some ideas and guidelines for managing vast amounts
of data.
Getting to the Basics
of NAS Storage
NAS is a method of presenting a shared area for many dif‑
ferent computers or users to store their files via a network.
Chapter 1

Creating a NAS system doesn’t take much more than these
components:
✓✓ A network, which is the key to being able to access data
from just about anywhere in a company as long as a com‑
munication link exists between the NAS server and your
clients’ computers.
✓✓ The physical storage devices, typically spinning disk drives,
that are located inside a server, also known as the filer.
✓✓ A user’s computer attached to the storage via the net‑
work that sees files and folders just like it does on a local
hard drive, except that this folder may have other users’
files in it as well.
One of the nice things about NAS is that depending on the
platform, tens of thousands of clients can be attached to the
shared storage areas, all happily reading and writing the files
they store there.
The ability to share data easily is fundamental to collabora‑
tion and efficiency in a workforce.
What made NAS grow so quickly and easily is that the basic
protocol, or language, that it uses to communicate isn’t pro‑
prietary, meaning that you don’t have to have a particular
operating system, server, or storage vendor to make it work.
The NAS protocol comes in two types, but both provide the
same feature set to allow shared access to files across a net‑
work from different client computers simultaneously:
✓✓ Network File System (NFS): This protocol was originated
and adopted as a standard on Unix‐based operating
systems.
✓✓ Server Message Block (SMB): This protocol came about
later. When Microsoft came out with the Windows oper‑
ating system, the company decided to adopt its own
network file protocol. SMB is widely referred to as CIFS
(Common Internet File System) in the NAS industry, and
for this reason, I use SMB/CIFS in this book to refer to
this protocol.
Both of these protocols are widely supported and formally
standardized across most operating systems and NAS ven‑
dors. A standardized protocol for NAS enables users to

Chapter 1: Understanding NAS 5
choose and share information without worrying about com‑
patibility issues with Linux, Windows, or Mac‐based client
operating systems. Users can select the best hardware and
software for their business.
Handling the Incredible Growth
and Management of Data
Data comes in two basic types: structured and unstructured.
Structured data has a definite format, such as what you would
find in a database. Each record, field, and data type expressed
within the file follows a predefined layout. What’s nice about
structured data is that you can describe every bit of the data
and explain what each byte of the file is. You can jump to
any spot inside a structured data file and by following the
predefined layout as if it were a map, you would know what
the bytes you’re looking at actually represent — for example,
a database that contains customer records with names,
addresses, account numbers, and most recent purchase
dates. You can, if you want, delete only the records of custom‑
ers who haven’t purchased anything for more than a year,
which would shrink the size of the database and save you
space. The structured layout makes it easy to find the records
you want to eliminate.
Unstructured data has no practical internal definition. Adding
or subtracting data from a file that is unstructured is much
more difficult than doing so in a structured file. You can do
it, but it isn’t as easy as simply running a database query to
add or delete certain records, as you can do with structured
data. Unstructured data usually takes the form of such things
as documents, spreadsheets, and data streams, such as audio,
photos, and video files. The application that created the file
knows how to deal with it, but in terms of the bulk management
of unstructured data, you are at the mercy of the sheer number
of possible applications that deal with these files correctly.
Over the last few years, the type of data created and the size
and amount of data that companies have kept on hand has
continued to skyrocket, and unstructured data is the biggest
offender in this growth (see Figure 1‑1).
You may have witnessed this yourself by building more and
more NAS systems to hold all this data. All these “NAS islands”

need to keep up with not only the physical storage of all this
stuff but also with the demand for client access. Apart from get‑
ting a call from someone complaining about a full NAS share or
slow performance, how do you gauge your storage efficiency
and limitations? I talk about that in the following section.
Finding an Efficient Solution
to Overloaded NAS Systems
When you have all that data to manage, you have to manage it
wisely. Data that was important yesterday may not be important
today — or it may be even more important now than it was then.
You never know how business needs will change the value of
your data, so dynamic data management is crucial to adapting to
changes. NAS storage has the great advantage when compared
to traditional block‐based storage of being more mobile and,
therefore, can provide intelligent data placement. The network
that NAS storage is attached to gives you that flexibility to move
your files around to wherever suits them best.
Having a layer of technology above all your NAS solutions that
recognizes access patterns to files and can shuffle and redis‑
tribute files according to changing business needs would be
an incredibly convenient tool to have. I talk about what that
solution is later in this book.
Figure 1-1: Unstructured data is growing more plentiful and bulky to deal with.

UnderstandingCloud
In This Chapter
▶▶ Discussing the cloud’s pluses
▶▶ Understanding that the cloud isn’t a perfect world
▶▶ Talking about the cloud vocabulary
▶▶ Looking at the differences between legacy computing and cloud
computing
▶▶ Explaining what makes up the cloud
The term “cloud” came about from IT folks doing diagrams
of their servers and networks to show how everything is
connected; when they didn’t know exactly what to draw for
a certain section, they drew a cloud‐shaped symbol. When
the Internet became part of the connectivity, IT people didn’t
know what “the Internet” equipment was either, so they drew
another squiggly bubble, resembling a cloud, to represent
“we’re connected to something inside this mysterious cloud.”
In other words, the cloud is just a metaphor for the Internet.
You don’t know what you’re talking to, but you know it’s there.
This chapter covers what cloud computing and cloud stor-
age are. It discusses the similarities and differences between
the cloud and the traditional server, storage, and network
infrastructures you may be accustomed to. It also goes into
the reduced cost of using cloud services and how you can
use the cloud to augment your current computing and stor-
age platforms securely and easily to save money, space, and
headaches.
Chapter 2

Appreciating the Cloud’s
Advantages
A significant reason that major companies have adopted
cloud is that the gear to run the cloud and store the data
doesn’t reside at a company’s physical site, which means that
the company isn’t building computer rooms, buying gear to
fill them, and maintaining that gear over its normal lifecycle.
Think about it like a rental car agency. Rental car agencies
buy the cars, house the cars, and maintain the cars. You just
pay them to use one whenever you need to.
Cloud works on a very similar idea. Cloud sites build the car,
and you rent the time to use it. When you’re finished and you
return the car, you pay for the time (and gas) and that’s all.
With the cloud, you can rent time on compute servers or rent
storage space so quickly, and on systems that are so fast, that
it would be impossible for your own staff to design, purchase,
and install your own storage and computing infrastructure
in the same amount of time. Cloud gives you that quick and
cost‐effective turnaround in setup, use, and eventual return of
computing resources.
Cloud services have another huge benefit. Because of the
scaled size of the resources that provide the computing and
cloud storage, the price you pay is incredibly cheap com-
pared to what you would pay to build it, power it, cool it, and
maintain it yourself. And because money is usually the driving
force in most IT decisions, and storing data inside the cloud is
very cost effective, this is actually the biggest cloud benefit.
Ah, Yes — the Cloud’s
Disadvantages
Every coin has two sides. As discussed in the preceding sec-
tion, the cloud has huge advantages. Unfortunately, though,
the cloud also has things that make people feel skittish. The
cloud is outside your normal data center walls, which means
that the gear that’s running it isn’t under your control. Also,
the immediate physical security and long‐term data storage
security is now someone else’s domain.

Chapter 2: Understanding Cloud 9
When someone else controls your storage and computing
environment, you may experience a little anxiety. However,
cloud service providers know that they have to deal with
that problem. Many, if not all, cloud providers protect their
environments by having no single point of failure, physically
and geographically. In other words, your applications living in
the cloud are actually capable of running anywhere and, most
likely, everywhere at some point. In cloud services, multiple
physical data centers that back up each other and protect the
physical equipment are the norm.
Getting your data to the cloud can be a challenge because the
cloud is far away. In networking terms, there is “latency” (or
delay) between your data center and the cloud. The latency
between your data center and the cloud can be 1/10 of a
second. This does not sound like much, but if you are pro-
cessing data in the compute cloud and this processing needs
to access thousands or even millions of files in your data
center, a million network delays of 1/10 second each really
add up and hamper productivity.
One thing that can’t be avoided is the speed of light. I recently
did a test from Indiana, connecting to one database in Virginia
and another in Oregon. For the database in Virginia, I got a
response time of 30 milliseconds. For Oregon, my response
time was 60 milliseconds. Welcome to latency, my friends.
No matter how speedy the equipment is in the cloud, data
still has to get from your location to the cloud. No amount of
clever engineering will let us go faster than light — the univer-
sal speed limit!
Another concern about the cloud that makes IT pros nervous
is data protection. Your organization is transmitting data out
of your data center and into the cloud provider’s network and
servers. Sure, security is in place to make sure that your data
is protected, but some people are a little queasy about not
having complete control over their own environments.
Remember that while your cloud vendor only secures the
physical aspects of the cloud, it’s up to you to secure your
data within the cloud.

Learning the Lingo:
Cloud Terminology
Cloud computing resources are, in essence, the same as those
of your typical data center:
✓✓ Servers that your applications run on
✓✓ Storage devices where your data is written to and
read from
✓✓ Networking resources that connect the servers, storage,
and users who exist outside the cloud to your applications
Pretty simple, right? The cloud isn’t that much of a mystery.
After you get the lingo down, you’ll be far more comfortable
with using the services that the cloud provides. Cloud servers
and networking are relatively the same with just a few nuances
thrown in, which I discuss later on. The major difference with
the cloud is how you store data, which is a good thing because
it makes things cheaper and much more flexible.
Storing data in a cloud, specifically with public cloud services
such as Amazon Web Services (AWS) Simple Storage Service
(S3) or Google Cloud Storage (GCS), part of Google Compute
Platform (GCP), is just one way of using the vast resources
of the Internet to store and retrieve objects. The reason I say
objects instead of files or data is because the concept of an
object is the atomic structure that you have to fully under-
stand before I can discuss other details of the cloud. But,
before I go into what objects are, you need to understand two
other terms: namespace and buckets.
Namespace
A namespace is a way to organizationally manage many
objects within a big container. Namespaces can span physical
locations and systems. A namespace contains uniquely named
items, such as buckets, which I talk about next. Namespaces
help to fence off a set of unique objects from other uniquely
known objects, which exist in other namespaces.

Think of a namespace as a universe. Each universe knows
about everything on its inside, but it doesn’t know or care
about anything outside its realm.
Buckets
Buckets are located within a namespace. A bucket, which I
think is a really great name, is a portion of a namespace that
is totally unique in name within a namespace.
Buckets contain objects, which are the equivalent of files. To
refer to an object, you need to know the namespace, bucket,
and object name, all of which are functionally equivalent to a
directory‐path/filename combination.
Objects
Objects are the actual data that you store and are all unique
items. Objects are addressed by something called a key‐value
pair, which has two halves, the first half of which is essentially
a filename, called the key‐name, that exists only once within
a globally defined namespace, inside a bucket. The value part
of the key‐value pair is the actual data within the object. For
example, here is how an object is referenced in AWS:
http://s3.amazonaws.com/bucket/key‐name
Google Cloud Platform (GCP) is similar. Here is how an object
is referenced in GCP:
http://storage.googleapis.com/bucket/
key‐name
Examining the AWS link more closely, it includes the following:
✓✓ http://s3.amazonaws.com: The web address of the
Amazon S3 namespace.
✓✓ bucket: Globally‐unique name for a bucket. You can
think of it as a root directory on a file system.
✓✓ key‐name: Unique name of an object stored in a
bucket. On a standard file system, this would be the
name of a file in a folder.

In a more detailed example, say that you have an Excel spread-
sheet that is your company’s (“XYZ Corp”) quarterly earnings
report. On your laptop, you might reference it this way:
C:XYZ_Corp_Data2015_Q2_Earnings_Report.xls
If you wanted to save that earnings report to the cloud, here’s
an example URL to reference that object in the cloud:
http://s3.amazonaws.com/XYZ_Corp_Data/2015_
Q2_Earnings_Report.xls
That’s pretty straightforward, isn’t it? XYZ_Corp_Data is
your bucket, and everything after the bucket is the object
(“2015_Q2_Earnings_Report.xls”).
What’s in a Name?
Here’s the catch. In standard file systems, you have directo-
ries and subdirectories (or folders) to help organize data in
some hierarchical fashion. Inside the cloud, you really don’t
have an exact concept of a folder or directory. Everything is
addressed with bucket/object names only, not bucket/folder/
sub‐folder/sub‐sub‐folder/object. Amazon S3 allows you to
cheat by using “/” in the key‐name of the object. So, for the
Excel spreadsheet example provided earlier, you can do this:
http://s3.amazonaws.com/XYZ_Corp_Data/
Reports/Public/2015/Q2/Earnings/Report.xls
In this URL, which is significantly more complicated, the
bucket is still XYZ_Corp_Data because that is always the
first thing after the namespace (s3.amazonaws.com). The
remainder is still the key to the object, which in this case is
the Excel spreadsheet for your 2015 Q2 Earnings Report. If
you go to Amazon S3 using their web‐browser‐based adminis-
tration interface and look inside the bucket, you’ll see folders
under the XYZ_Corp_Data bucket that you can click on to
drill down inside them like in a normal directory or folder,
until you get to the actual “Report.xls” object. However,
the key‐pair for that file is, and always will be, “Reports/
Public/2015/Q2/Earnings/Report.xls.” You can’t refer
to it only as “Report.xls” in any sense because http://
s3.amazonaws.com/XYZ_Corp_Data/Report.xls and

the original URL refer to two different bucket/key‐pair names
entirely.
Be careful how you think of traditional filenames and object
names within the cloud. No more is there a referential view of
being inside a folder and then pointing to the files inside that
folder and only that folder. It’s also important to remember
that there is no concept of a directory or folder, as in a stan-
dard file system, and an object doesn’t actually exist inside or
under a folder. Amazon S3 allows you to pretend that a folder
structure exists, but there really isn’t any such concept. The
object’s key‐name is a fully qualified name from the bucket
name down, and you can’t rename these folder names within
the key‐name and expect any other object that shares that
same folder name to also magically move to the new folder —
because there isn’t a folder; what looks to be a folder name is
actually part of the object’s name. You would have to rename
every object’s key‐name one by one in order to change the
pseudo directory/folder representation.
You must be specific every time you describe an object, from
the bucket all the way down to the object key‐pair names. You
can translate your current methods of storing and retrieving
data in the manner you’re accustomed to into the cloud’s
object‐based storage scheme. Avere Hybrid Cloud NAS does
this transparently for you as part of their data‐tiering capabili-
ties when it stores data in the cloud.
Parts of the Cloud
A misconception exists that the type of equipment that makes
up the cloud is different from standard computer gear that
you would use in your own company’s data center. This isn’t
the case. The cloud uses the same server, storage, and net-
working gear that any regular IT shop uses — but it’s the way
it uses them, and what they do, that makes the difference.
The basic cloud ingredients are
✓✓ Compute servers, which run an OS such as Windows,
Linux, and so on
✓✓ Storage servers, which hold your files
✓✓ A network, for communication

Compute servers
Compute servers are where the application magic happens.
These servers read the data, crunch the data, and then spit
out information. Servers in the cloud tend to be grouped into
a compute farm or a multiserver pool of resources consisting
of CPU, memory, and sometimes limited local disk storage.
This pool is “virtualized” into one super‐server, and a virtual
server is created by slicing off X number of CPUs, Y number of
memory, and Z number of virtual disks.
Many of these virtual farms are running VMware or Xen virtu-
alizing software to do all the slicing and dicing of resources.
Think of a big loaf of bread being the pool of all the servers’
CPUs, memory, and disks, with a slice of bread from the loaf
as your new virtual server that behaves like a single, physi-
cal server. Making all the resources of the server completely
virtual allows a system administrator to move the operating
system and applications from one physical server to another
on the fly, without interruption.
When you request a server to be created in the cloud, you can
describe what operating system you want it to run, how much
memory you need, and the CPU count. You also can choose
from varying types of standard setups, depending on what
kind of applications you’re planning to run.
Storage
Although it can vary, cloud storage is always a type of shared
storage that can be connected to multiple servers simultane-
ously. This ability to share makes it easy to slice off what you
need and give it to whichever server needs it. When you’re
done, you can even disconnect it and give it back to the pool
for other servers to use again.
Storage in the cloud can have different characteristics, such
as redundancy and a guaranteed I/O rate. You’ll pay more
per gigabyte for highly available and/or faster storage, which
is a choice you make depending on what you need when you
request the storage resources through the cloud administra-
tion tools.

Network
The glue that makes everything work is the network. Cloud
networks are virtual, and they’re great because you don’t have
to run around plugging and unplugging wires to get servers
and storage and clients to talk to one other. The servers are
connected to the network when they’re built, and everything
else, such as IP addresses, routers, DNS entries, and firewall
rules are all handled either automatically or customized with
management software by the administrator.
Looking at Cloud Types
Clouds come in two basic “flavors.” You can build your own
internally functioning cloud, which is called a private cloud,
or you can use a publicly accessible cloud solution called a
public cloud. The cloud also has a third type, but instead of
a flavor, it’s more of a mix known as the hybrid cloud, which
is just a private cloud and a public cloud connected together
that pass data back and forth as needed.
Private clouds
Private clouds are storage and computing environments
that you would build yourself, except that all the interaction
between them and your applications would happen in “cloud-
ese,” or the protocols that cloud computing is accustomed to
speaking. For example, you can use TCP/IP addresses exclu-
sively to communicate with applications, use object‐based
references to data, as discussed previously, and have web‐
browser‐friendly, client‐side applications.
Private clouds function in the same way as public clouds
except that all the hardware is inside your own data center
and not out on the Internet being shared with everyone else.
A private cloud places everything under your control, but the
downside is that you have to build it and maintain it.
Public clouds
Public clouds, which are the more prominent cloud services
solution, exist in the Internet’s shared space. When so many

different companies share the resources, the price of compute
and storage on the public cloud goes way down. You pay only
for what you use, which is perfect for companies that have
sudden demands for computing resources such as a product
launch, a blockbuster movie to complete, or a big genomic
study. When the demand for resources goes down, or is no
longer necessary, the resources are returned to the public
cloud for others to utilize, and you no longer pay for them.
Hybrid clouds
You don’t have to pick one type of cloud over the other
because the hybrid cloud is a mix of using both private and
public cloud services together, which gives you the control
of in‐house private cloud resources plus the cost savings and
flexibility of public cloud.
Hybrid clouds can involve using private cloud to do day‐to‐
day data processing and then storing that data to the public
cloud after it becomes less useful but is still important enough
to keep. If you use the data less frequently, then sending it
out to the public cloud is usually a more cost‐effective way of
storing it for the long term.
Hybrid clouds can also be used in the reverse fashion, where
processing is done in the compute cloud and data is stored
in the private cloud in the company’s data center. This cloud
bursting approach takes advantage of the vast compute
resources available in the public cloud and can alleviate
security concerns of moving your data storage to the public
cloud. Cloud bursting is discussed in detail in later chapters
of this book.
Moving data from within the company’s private cloud to the
public cloud isn’t a trivial operation, but using Avere Hybrid
Cloud NAS to bridge the gap between them is an efficient way
to do so.
Accessing and Securing
the Cloud
How you communicate with cloud resources is simple. If
you’re using a private cloud inside your own corporate

network, the cloud appears in the same way that any other
application server does. A public cloud solution isn’t that
much trickier.
After you have TCP/IP connectivity, the cloud needs to con-
firm your identity before allowing you to access your servers,
namespaces, buckets, and objects. Security is tightly con-
trolled in every public cloud solution.
Speaking generally about most cloud security, you’ll have spe-
cific operations that can be performed on certain components
within the cloud. These abilities or roles are usually assigned
or revoked by a user or a group of users. Each user has a
specific set of authentication/access keys that have to be veri-
fied every time they communicate with the cloud provider.
These keys ensure that the application (via a user interaction)
is a legitimate cloud services user and can perform the tasks
requested by the application. Basically, the security of cloud
resources is as good as, if not better than, your current inter-
nal IT infrastructure. Security is everywhere in cloud, from the
access to the servers to the communications between them
and all the way down to the objects placed in the buckets.
Everything can be controlled and monitored.
By using a cloud solution such as Avere Hybrid Cloud NAS, in
addition to normal user‐ or group‐based access, you also have
the ability to encrypt the data that is stored on the public
or private cloud. Doing this ensures that nobody can access
the objects (files) you’ve stored there, so they’re safe from
unauthorized eyes. Avere Edge filers can decrypt and access
the data held within, which helps to ease the worry of storing
important proprietary customer data on the Internet.
Transitioning into the Cloud
Now, you’re probably thinking, “Okay, so how do I make this
happen without throwing out my existing gear?” Don’t worry.
The Avere Hybrid Cloud NAS solution will help you optimize
the investment in your current NAS environment while gaining
all the benefits of cloud discussed here. Chapter 3 explains
exactly how to bridge the gap between the cloud and NAS to
create hybrid cloud NAS.

IntroducingAvereHybrid
CloudNAS
In This Chapter
▶▶ Linking your on‐premises storage with cloud storage
▶▶ Unlocking the power of the cloud
▶▶ Connecting cloud compute with local storage
▶▶ Activating the hybrid cloud
In Chapters 1 and 2, I talk about NAS and cloud storage,
respectively. This chapter explores Avere’s innovative
physical and virtual Edge filer technology. You will learn how
you can use these advanced pieces of engineering to tap into
the power of cloud computing and cost‐effective storage while
maximizing your investment in on‐premises infrastructure. By
combining local and cloud infrastructure into a hybrid envi‑
ronment, you get the best of both worlds.
Bringing NAS and the Cloud
Together As One
Enterprise companies are used to providing business value
through the delivery of IT services from physical data centers
and co‐location facilities. IT departments face ongoing budget
pressures and increasing service demands. For departments
accustomed to thinking of data centers in terms of megawatts,
the rise of the public cloud represents an entirely new, cost‐
effective toolset for both storage and compute.
Chapter 3

Traditional companies have an architecture made up of local
compute servers talking to a local NAS. With latency concerns
hanging over their heads, IT directors and storage architects
look longingly at the cost‐effective yet unfamiliar object stor‑
age provided by vendors like AWS and Google Cloud Platform,
which offer Simple Storage Service (S3) and Google Cloud
Storage respectively. Staring at an existing on‐premises NAS,
IT pros like you want nothing more than to take advantage
of cheap cloud storage while maintaining local performance.
This is where Avere comes in with its FXT Edge filer series.
In this book, I focus on AWS and Google Cloud Platform
because these are the cloud providers that Avere Hybrid Cloud
NAS supports today. Other providers may be supported in the
future and the same capabilities would apply to them.
As shown in Figure 3‑1, an Avere FXT Edge filer is a physical
appliance that is set up between your on‐premises compute
and storage environments. Filled with intelligent software that
can cache and tier data, loaded with gigabytes of RAM and
terabytes of SSDs, Edge filers provide your compute environ‑
ment with fast access to storage. Smart caching and pre‐fetch
algorithms ensure that active data is kept as close as pos‑
sible to compute to help optimize application performance.
Capable of being deployed in clusters from 3 to 50 nodes,
FXTs can be horizontally scaled to provide millions of IOPS.
An FXT cluster can also hook into the public cloud and pro‑
vide on‐premises compute with a global namespace that
spans both local and cloud‐based storage. Using Application
Programming Interfaces (APIs) to connect to local or cloud
object storage, the FXT can present both NFS and SMB/CIFS
to your compute environment. With its algorithms and fast
media doing the heavy lifting, the FXT cluster addresses the
latency problem typically associated with the public cloud.
With an appropriately‐sized cluster, latency to cloud storage
becomes ancient history.
Interested in instant geographic diversity and disaster recov‑
ery for your local NAS? I thought so. Baked into the FXT is a
feature called FlashMirror®
, which can mirror data between
local and cloud storage environments. Trying to migrate
from a local NAS to the cloud? Storage engineers can use
FlashMove®
, a data mobility tool built into the FXT, to non‐
disruptively migrate data between local and cloud storage.

Chapter 3: Introducing Avere Hybrid Cloud NAS 21
That all sounds well and good, but what if you are running
compute workloads in the public cloud and need to access
data housed in on‐premises storage? That’s where Avere’s
Virtual Edge filer (vFXT) comes into play.
vFXTs perform just like their physical brothers, just in the
opposite direction. Loaded with the same software, vFXTs can
also be deployed in clusters from 3 to 50 nodes, giving them
identical scaling characteristics. vFXTs also have the same
built‐in data management tools as the physical FXTs, giving
your storage team a familiar interface.
Capable of being deployed on AWS and Google Cloud
Platform, vFXTs are configured on memory‐optimized
instances and connected to SSD‐based block storage. The
same algorithms that power the physical FXT allow your
cloud compute environment to have global namespace visibil‑
ity to both cloud storage and on‐premises NAS. This lets you
worry less about where the data resides today.
Figure 3-1: Avere FXT Edge filers provide a global namespace that spans
local and cloud storage.

A vFXT cluster frees you from storage management complex‑
ity. Instead of fretting over the IOPS that the public cloud can
provide, use General Purpose Elastic Block Store (EBS) on
AWS or Persistent SSD in Google Cloud Platform. Simply size
your vFXT cluster to get the performance you need.
Things get really interesting when you consider the ultimate
in hybrid cloud flexibility, as shown in Figure 3‑2.
By combining cloud‐based virtual FXT and on‐premises physi‑
cal FXT Edge appliances, companies have the flexibility to
deploy applications wherever it makes the most sense. With
compute‐to‐storage latency no longer a concern, other factors
like capital versus operational expense, equipment lifecycle,
geographic data diversity, and cost can take priority.
Empower your people and yourself to think beyond tradi‑
tional latency constraints. Treating latency as a solved prob‑
lem allows your storage team to develop creative solutions
for delivering business value without breaking the bank.
Figure 3-2: Physical FXT + Virtual FXT = Ultimate Flexibility.

Using the Best of Both
NAS and the Cloud
The best teams are made up of people with differing perspec‑
tives and points of view. Similarly, combining physical and
virtual FXT Edge filers as components of your overall storage
architecture allows you to maximize the best features of local
NAS and cloud‐based storage. This lets you solve storage chal‑
lenges that can’t be independently fixed by either traditional
NAS or cloud storage. Speed, latency, cost, and complexity
are the main drivers in managing storage. By combining the
features that make NAS or cloud a good solution on their own,
you can creatively deliver business value in ways that previ‑
ously were not possible.
Making things faster
Speed! We all crave it. Humans have been obsessed with
speed since the dawn of time. If it can move, people end up
seeing how fast it can go: bicycles, motorcycles, cars, yachts,
planes. Let’s explore how you can optimize storage perfor‑
mance for your local or cloud compute environment using
Avere’s Hybrid Cloud Edge‐Core architecture, no matter
where your compute or storage sit.
The ability to use both physical and virtual FXTs ensures
speedy storage performance for your compute farm.
Separating performance from capacity, Avere’s Hybrid Cloud
solution delivers both more efficiently. The Edge filers supply
the performance your cloud or local applications need,
abstracting the slower performance of local or remote stor‑
age, as illustrated in Figure 3‑3.
Figure 3-3: Avere Hybrid Cloud Edge‐Core architecture.

The Edge filer provides high‐performance and low‐latency
data access to the compute nodes. This is possible due to
three distinctive features of the FXT and vFXT:
✓✓ Dynamic caching, pre‐fetching, and tiering algorithms run
inside both FXT and vFXT Edge filers to automatically
move the hottest data blocks from the slower disks of the
core filer(s) to the Edge filers’ fast media (RAM, SSD, and
in the case of the physical appliance, SAS).
✓✓ As a full‐featured file server, both virtual and physical
Edge filers locally terminate all read, write, and metadata
operations. Write data is asynchronously written back to
the core filer(s) on a schedule you set, making latency a
thing of the past.
✓✓ Clusters of Edge filers, from a minimum of 3 and going up
to 50, work together as a group, distributing file requests
from clients and dynamically scaling the I/O workload
across themselves.
With an Edge cluster on the front lines providing hot data,
the demands on the core filers, both on and off premises, are
reduced. This allows for better optimization of both capacity
and cost savings.
Making things cheaper
The explosive growth of unstructured and structured data is
placing immense pressure on storage costs. The public cloud
offers the promise of financial relief in the form of cheap, pay‐
as‐you‐go storage. Hybrid cloud architecture lets you realize
that potential savings no matter where your applications and
compute resources live.
While data has value, its storage comes at a cost. If you look
at the number of times a piece of data is accessed, you can get
a good sense of how valuable it is to your business. By using
algorithms to calculate the usefulness of data based on activity
levels, both vFXT and FXT Edge filers know how to accurately
figure out the value of a specific piece of data. The Edge filers
let you save money by automatically moving low‐value, infre‑
quently accessed data to the cheapest storage location — on
premises or in the cloud. Because this happens automatically,
the Edge filers are constantly working on making sure that the
most valuable, frequently‐accessed data is closest to the sys‑
tems that need it.

Once comfortable with the concept of using Edge filers as a
buffer to meet the I/O demands of mission‐critical applica‑
tions, you can turn your attention to cutting the cost dragon
down to size. What’s the cheapest place for bulk storage for
your company? Local NAS? AWS? Google Cloud Platform?
By sizing your Edge cluster appropriately, you can take
advantage of the operationally cheapest long‐term storage
regardless of location, while having the confidence that your
applications will meet their performance targets. If one public
cloud becomes significantly cheaper than another, you can
rest easy knowing that switching long‐term storage vendors
won’t impact your line‐of‐business applications.
Adapting existing applications to use object storage is expen‑
sive. Adapting vendor applications to use object storage is
out of your control. Edge filers let you take advantage of cloud
object storage while presenting familiar NFS and SMB/CIFS
interfaces to the applications you rely on. This means your
application teams can spend more time working on the fea‑
tures that uniquely give your business an edge, and less time
worrying about storage interface protocols. Without the need
for sweeping application architecture changes, Edge filers
give you flexibility and help your bottom line.
Using the public cloud and concerned about IOPS? Do the
math and consider putting a vFXT cluster in front of AWS S3
or Google Cloud Storage. No need to sweat the cost of putting
all your data in block storage on AWS EC2 or Google Compute
Engine — just front AWS S3 or Google Cloud Storage with a
vFXT cluster and scale it as you need to!
If you come to the common conclusion that it is more
cost‐effective for your business to let AWS or Google Cloud
Platform operate your physical storage infrastructure, Avere’s
vFXT and FXT Edge filers give you the performance your
applications demand while letting you choose the cheapest
storage backend.
Making things simpler
Both physical and virtual FXT Edge filers offer the Avere
Global Namespace (GNS), simplifying data management and
access in a geographically diverse storage environment.
Avere’s GNS spans public object, private object, and legacy
NAS, all from heterogeneous vendors (see Figure 3‑4).

Need to access data from Google Cloud Storage with your
local Red Hat environment? No problem, your physical FXT
cluster handles the object‐NFS translation. What about data
locked up in your local NAS that you need to get to from your
AWS Amazon Linux environment? This is a challenge that
many organizations face as part of the cloud transition. With
Avere’s FXT‐powered GNS, compute clients have a single
mount point on the FXT cluster and have access to all the
data, regardless of where it is ultimately stored.
Avere’s GNS implementation doesn’t remap your existing file
systems. No need to worry about vendor lock‐in, because the
GNS is simply an overlay of the namespace you create.
With the GNS in place, you have a single map to all your
data repositories. You can transparently migrate data using
FlashMove, a data management tool within both the virtual
and physical FXT filers.
Easy Data Migration
Avere FlashMove is a piece of data management software that
runs on both virtual and physical FXT Edge clusters. Available
wherever an FXT/vFXT filer is, this software makes optimizing
heterogeneous storage environments a breeze.
Figure 3-4: Public object, private object, and legacy NAS in a global
namespace.

Thinking of switching local NAS vendors? No problem! Since
Avere supports multiple legacy NAS vendors including EMC
and NetApp, migrating from one to the other is as simple as
executing a FlashMove command. Want to trade power, cool‑
ing, screwdrivers, and racks for the public cloud? FXT/vFXT
Edge filers can FlashMove your data for you without changing
the storage interface for your applications. No application
rework required. Easy!
As sure as the Earth revolves around the Sun, your storage
team will have to perform data migrations. There are many
causes, including hardware lifecycle and switching NAS
vendors. FlashMove eases the pain on your storage team by
reducing the effort required, transparently and non‐disruptively
moving data from your original storage subsystem to your
new storage target. Users and applications continue to see
live data while FlashMove handles data migration in the
background.
With FlashMove, you don’t need to halt applications or sus‑
pend access to data during migrations. FXT and vFXT clusters
continue serving active data to application servers and users
while the FlashMove software transparently migrates data from
source to target storage subsystem. You can load‐balance live
data across existing systems, transparently archive to slower/
cheaper storage, decommission old storage, add new storage
vendors, and take advantage of public cloud storage all with a
few simple commands.
Every FlashMove can be done on live data without disrupting
the operations of your existing applications! Non‐disruptive
NAS to cloud, cloud to NAS, or cloud to cloud data migrations
are at your fingertips.
Protecting Your Data
Running a data center is an expensive proposition. Losing the
data on which your company relies is an unacceptable propo‑
sition. Nobody wants to tell the CEO about data loss. Making
copies within a single geographic area won’t help when a
natural disaster rolls in — just ask any company that had
its operations impacted by Hurricane Sandy in 2012. Having
physical access to your data center doesn’t matter if that data
center is literally under water.

Building out a data center in a separate part of the country is
cost‐prohibitive for most companies. Co‐location facilities are
only slightly cheaper. Frequently, the public cloud holds the
most cost‐effective solution to this challenge, offering cheap
object storage in multiple geographic regions across the
globe. While you yearn to take advantage of this capability,
Avere helps you get there by building FlashMirror into every
FXT and vFXT Edge filer.
Let’s say you’re happy with the cost and performance of your
local NAS, but find your current configuration doesn’t meet
your organization’s recovery time objectives (RTOs) or recov‑
ery point objectives (RPOs). With FlashMirror, your FXT Edge
filer comprehensively protects your critical data by mirroring
it in either AWS or Google Cloud Platform. Instead of incurring
the cost of establishing multiple physical remote data centers,
you can rely on the public cloud. After all, AWS and Google
Cloud Platform employ seismologists when selecting sites for
their respective data centers.
When considering cloud object storage, you can’t compare
apples and oranges. Every time you write to S3, the data you
write is placed in at least three different data centers. Can you
say that about your local NAS?
From the moment you execute the FlashMirror command
against a source directory, your FXT cluster creates a base‑
line copy in secondary storage of the source directory. Like a
legacy NAS mirror, FlashMirror ensures that any change made
in the primary storage is reflected in the secondary location.
Prior to the next scheduled Disaster Recovery (DR) review at
your company, talk with your storage architects about how
FlashMirror can help achieve the RTOs/RPOs management is
asking for, improving your data protection and providing for
business continuity without exorbitant additional expense.
Connecting to the Cloud
The public cloud offers practically infinite, pay‐as‐you‐go
compute and storage capacity. With such massive economies
of scale and the possibility of limiting waste, traditional com‑
panies are rapidly developing plans to take advantage of the
capabilities the cloud offers. The challenge is that most cloud

storage is object storage — how do your existing applications
connect to it?
Avere FlashCloud™ provides the one answer to this question.
As a feature of both vFXT and FXT Edge filers, FlashCloud
integrates object storage with legacy NAS into a global
namespace, providing scalable performance for users every‑
where via familiar NAS protocols.
GNS lets enterprises adopt object storage at the pace of their
choosing, while offering the flexibility to store data wherever it
makes most sense. FlashCloud is the piece of technology that
abstracts object storage interaction by providing enterprise‐
class NAS functionality protocols, including NFS and SMB/
CIFS. Working with cloud and local storage, vFXT and FXT
filers give you optimized performance, capacity scaling, and
lower total cost of ownership.
Many companies get started in the cloud by simply hooking
up a physical FXT cluster to cloud storage. This is a well‐
travelled path, offering cost‐effective alternatives to tradi‑
tional NAS environments. What makes the Avere Hybrid Cloud
more interesting is when you consider the power of opening
up your on‐premises storage to compute resources in the
cloud.
Across many industries, it is commonplace to have nightly
batch jobs that act upon a tremendous amount of data. In
order to accommodate this resource‐intensive processing,
application teams in traditional companies frequently have to
over‐provision their local compute resources. A more cost‐
effective alternative is to present existing local storage to the
cloud using a vFXT cluster. Remember, in the cloud you only
pay for what you use, which means you can intentionally use
transient servers as part of your architecture. These servers
don’t have to persist — they can simply be instantiated, per‑
form a data processing task, and be terminated. Cloud burst‑
ing at its finest!
With Avere vFXT Edge filers, companies can fire up compute
resources on AWS or Google Compute Platform and simply
point them to the filer cluster. Without moving the source
data, companies can burst cloud compute resources at will.
This gives companies the freedom to select cloud compute
vendors based on cost.

Each company has different storage and compute require‑
ments, objectives, and goals. Properly designed, the hybrid
cloud lets you maximize your existing on‐premises invest‑
ments while seamlessly letting you take advantage of the
public cloud. Just upgraded a massive legacy NAS? No wor‑
ries, you can still optimize cost using cloud compute.
Making Things Better
Faster, cheaper, more robust, scalable, and easier to
operate — that’s my definition of making things better. Let’s
explore robustness and scalability for a moment. Traditional
design practice to address these issues is with a highly avail‑
able cluster. In keeping with industry‐accepted best practice,
the vFXT and FXT Edge filers can be clustered in a high‐
availability configuration. This helps ensure that access to
storage is resilient, robust, and scalable.
Hybrid cloud provides the flexibility to maximize existing cap‑
ital investments while taking advantage of variable compute
and storage in the cloud. This lets you focus precious capital
dollars on improving the robustness and reliability of archi‑
tectural components that aren’t candidates for moving to the
cloud. This flexibility also lets you adopt the cloud at a pace
that is right for your organization.
There now, isn’t that better?
Hybrid cloud combines the well‐understood connectivity and
ease‐of‐use of a legacy NAS with the future direction of data
storage. Scalability, investment protection, and consolidated
management gives your organization the infrastructure it needs
so you can spend more time focused on delivering value.

ReapingtheBenefits
ofCloudBursting
In This Chapter
▶▶ Exploring benefits
▶▶ Noodling over usage models
▶▶ Zero constraints
▶▶ Operational excellence
In this chapter, I explore many of the non‐technical ways in
which companies can benefit from cloud bursting. You’ll
see that the ability to cloud burst truly changes the paradigm
of how to think about compute and storage.
Benefits
Technology is flashy and fun, but it doesn’t mean a thing
if your cash register doesn’t ring. Ultimately, IT is about
enabling the business to create and deliver value. Let’s
explore some of the non‐techie benefits of cloud bursting.
Faster time to market
One path to maintaining your competitive edge is by shorten‑
ing cycle times. The faster your company reacts to changes
in customer demand, market forces, or regulatory con‑
straints, the better positioned you are for sustained success.
Sometimes, thousands of hours of processing time can stand
between you and the next release of your product. This is
Chapter 4

true in many industries, whether producing the next smash
hit movie, designing the next cancer‐fighting drug, crunching
research data, or developing the next race‐winning car. If your
project is running behind schedule, the hybrid cloud can give
you an edge.
We know that the cloud has almost limitless compute capac‑
ity. With a few mouse clicks or a bit of scripting, you can
spin up thousands or tens of thousands of servers on AWS or
Google Cloud Platform to add to the compute power you have
available. With the ability to scale storage access to keep pace
with this augmented compute power, your project will be
back on track.
Being early is definitely better than being late. Incorporate
the ability to cloud burst into your product planning cycle.
Use the knowledge that you can bring tens of thousands of
machines online to improve your organizational agility.
Pay only for what you use
So you’ve saved the day by bringing a project back on sched‑
ule by cloud bursting. Congratulations. You know what you
can’t forget to do after you’ve processed that mountain range
of data? Shut those servers down, just like shutting off the
lights when you leave your office. Remember, in the public
cloud, you only pay for what you use. This lets you dial in
exactly the right amount of performance when you need it.
Using a hybrid cloud approach, you have the added advan‑
tage of dialing in performance where you need it.
Focus on your business, not IT
In most companies, IT is not the core strength. IT exists to
serve the strategic objectives of the business. Sometimes, IT
pros get so caught up in the joys of tech that they forget: The
entire point of IT is to serve the mission of the company.
As compute and object storage becomes increasingly com‑
moditized, AWS and Google Cloud Platform are able to realize
massive economies of scale. If you think your company is in
a position to design a more efficient, more cost effective data
center, you should seriously consider getting into the public
cloud game.

Chapter 4: Reaping the Benefits of Cloud Bursting 33
Joking aside, the point is that global data center consolidation
is happening. Let the pros at AWS or Google Cloud Platform
handle racking, stacking, power, and cooling. You can still
access your data when and where you need it via the hybrid
cloud.
Letting go of the hardware lets you focus on your core mis‑
sion. Go ahead, unleash your artistic creativity for that next
movie. Focus on the science required to achieve that medi‑
cal breakthrough. Optimize your fueling strategy to win that
championship. And rest well at night, knowing that by adopt‑
ing a hybrid cloud infrastructure now, you can ensure that the
data on which your business relies is accessible, regardless of
its physical location. This lets your techies plan your future‐
state architecture to meet your business’ timelines.
Improve your financial agility
By using the hybrid cloud, you don’t really care if AWS or
Google Cloud Platform is providing the back‐end object store.
Go ahead, let these two technical juggernauts slug it out for
global domination. With a storage cluster as a layer in front of
your end‐state storage, if Google Cloud Platform or AWS pro‑
vides a significant cost advantage over the other, all you need
to do is migrate your data to the cheaper provider. Courtesy
of the hybrid cloud, we’re in a world where financial engineer‑
ing is becoming as important as computer engineering.
Usage Models
There are a couple of ways companies are approaching cloud
computing. Let’s explore them together.
Temporary usage at peak times
Consider the following questions: Is your business affected by
seasonality? Are your demand swings predictable or spiky? If
you answered yes, your organization can benefit from cloud
bursting. As we’ve discussed, you only pay for cloud compute
or storage when you are using it. Cloud burst as the variability
happens to give your customers the performance they expect.

What about non‐production systems? Let’s do some simplified
math. Say you have an equal number of development, test,
and production systems. If you can turn off your development
and test systems while your staff is peacefully slumbering, the
savings add up quickly. Embrace cloud bursting by turning
these systems off when they are not being used.
Permanent usage, full time
Are you beyond the point of temporary usage? Are you ready
to turn in your wrenches and screwdrivers and commit to a
cloud‐centric compute environment? Do you want to migrate
to the cloud but run into snags, hampered by legacy applica‑
tions that don’t understand object storage? The hybrid cloud
facilitates this transition by allowing you to move your data
where and when you are ready, with the storage protocols you
already use, at a pace you control. Think about it — you don’t
need to make changes to your application stack to start realiz‑
ing the numerous benefits of cloud computing and storage.
Data Can Be Anywhere
For traditional companies, most data lives on premises today.
The trend is for that data to migrate to cloud storage in the
years to come. Commonly, applications are hamstrung by data
location. Migration strategies hinge on data location, with some
applications unable to move until a database moves.
With a hybrid cloud design, your cloud‐based applications
can access data from anywhere, at NAS speed, even if data
still resides on premises. While taking advantage of this unique
design pattern, you can start planning when to migrate your
data to complete your transition to the cloud. Meanwhile, you
can offload your cold on‐premises data and start using cloud
storage today.
Speed of Flash, Cost of Disk
There’s a saying in racing, “Speed costs money — how fast do
you want to go?” For your applications to perform fast enough
to meet the demands of your business, you need to ensure
that your compute has consistently fast access to storage.

Chapter 4: Reaping the Benefits of Cloud Bursting 35
This level of performance is possible using a physical appli‑
ance in front of cloud storage or a virtual appliance in front
of on‐premises storage. Thus, rarely‐accessed data can be
offloaded by the appliances onto whatever storage is most
cost‐effective.
Okay, so cloud object storage is cheap, but what about the
speed part of the racing question? That’s where you have to
rely on the software within physical and virtual filers. With
intelligent algorithms, hot files are identified before you even
know they exist and are automatically pulled into the filers’
RAM. As the file cools off, it is automatically tiered to progres‑
sively slower media, passing through SSDs before becoming a
tiny ice cube in the corner of the cheapest object storage you
can find.
If a file ever looks like it’s going to get hot again, pre‐fetch
software comes into play, pulling the file back up the storage
tiers, so your applications never know the difference or expe‑
rience a dip in performance.
If your car didn’t have gauges, you wouldn’t know how fast
you are going or how much gas you have. Similarly, when
evaluating solutions for cloud bursting, don’t underestimate
the importance of a solid dashboard to give you a perspec‑
tive into how your system is performing. On the storage side,
make sure that you can see hot clients, files, IOPS trends,
throughput, and latency. This gives you the confidence to
know that your applications are benefiting from the speed of
flash, while you pay for the cost of disk.
Scale, Baby, Scale
Time and budget are not the only criteria by which a suc‑
cessful IT project is measured. If you are on time and under
budget, but your solution doesn’t scale, have you achieved
success? Only if you don’t plan to serve many customers or
to be ready for the increasing demand that inevitably comes
from the explosion of data. Using a hybrid cloud solution that
has been designed for growth means you’re ready for next
year’s challenges when they come.
Let’s say you’ve built your infrastructure to produce one movie
or to develop one drug. As your business succeeds, you want

to be able to easily grow your IT infrastructure to support
working on two movies or two drugs simultaneously. That’s
what the clustering capability of the Avere Hybrid Cloud deliv‑
ers. Simply add nodes, without downtime, to the Avere cluster
and keep pace with the progress of your business.
With the ability to scale both vFXT and FXT clusters to a maxi‑
mum of 50 nodes, Avere is designed to accommodate scale.
The Standard Performance Evaluation Corporation (SPEC) is a
non‐profit corporation that develops standardized benchmarks.
Avere pitted a 44‐node FXT Edge cluster against Solution File
Server (SFS), one of these benchmarks. The result? According
to data publicly available and peer‐reviewed at https://www.
spec.org, the Edge cluster delivered 1.56 million IOPS with an
overall response time of 0.99 milliseconds. Multiple results have
been posted to the SPEC website by Avere, including several
where Avere was configured with cloud storage. Check it out!
For those of you with an unrelenting desire for details, check
out Figure 4‑1 for a delicious configuration diagram of the
44-node Avere test cluster that achieved 1.56 million IOPS!
Scale? Oh yes, this storage system can scale. Bring on your
high performance computing (HPC) and big data workloads!
Figure 4-1: Avere delivers record-setting SPEC SFS performance.

LookingatCloudBursting
UseCases
In This Chapter
▶▶ Lights, Camera, Action!
▶▶ Understanding market forces
▶▶ Combatting cancer
▶▶ Achieving your wildest imaginings
In this chapter I explore ways in which cloud bursting
applies across a wide variety of industries. Feel free to skip
around to the use case that is closest to your heart, or go in
sequence to see how different challenges are being tackled.
Cloud Bursting Is Everywhere
The easy access to virtually limitless computing is funda-
mentally changing the way companies can approach prob-
lems. Animation and visual effects studios burst to massive
compute farms, creating spectacular movies. Educational
institutions burst to the cloud, slashing response time to con-
stituents around the world. When medical researchers need
to perform genomic research, they turn to the cloud. Retail
operations burst to the cloud, dealing with seasonal trends.
Entertainment companies use cloud bursting to scale up and
support wildly popular video streams.
Let’s dive into some examples and explore the challenges
faced by some specific companies, and how they’ve used
cloud bursting to solve those challenges.
Chapter 5

Movie Rendering
The creation of any modern movie requires an incredible
amount of computing power. Individual frames can often
take hours to render, with some films consuming hundreds of
terabytes of storage. That’s a lot of data! Leading visual effects
companies have built massive compute farms in excess of
40,000 cores. To ease our collective appetite for even more
amazing films, the case for cloud bursting is clear.
Visual effects companies have made massive investments in
infrastructure. This means these companies are operating
their own data centers, frequently in places where real estate
square footage commands a premium. Cloud bursting allows
these companies to take maximum advantage of their existing
infrastructure investments. There is no need to rush to the
cloud and short‐circuit the hardware lifecycle. Instead, the
IT pros can focus on keeping their compute farm running at
close to one hundred percent.
You may not know that the visual effects business is seasonal,
just like retail. There is a big push around the end of year
and beginning of summer to get movies into theaters. These
demand surges often eclipse existing render farm capacity.
When that happens, you can’t build out your data center
overnight. It is cost prohibitive to move all of the raw data
that needs to be rendered into the cloud. And you can’t attach
cloud compute directly to a local NAS, because latency to
even the closest cloud data center ranges between 10 and
100 milliseconds. With that much latency, storage performance
is simply unacceptable. Fortunately, there is a better solution.
With the ability to deploy a virtual filer cluster in the cloud,
demand spikes can be effectively managed. Simply spin up a
storage cluster on AWS or Google Cloud Platform, point it to
on‐premises storage, and let the algorithms work their magic.
With the ability to bring thousands of additional cores to bear
and with latency no longer a concern due to filer caching,
visual effects companies can make sure they meet rendering
commitments on time. An additional benefit you get from filer
caching is a decrease in the overall Wide Area Network (WAN)
load, which shrinks networking costs.
This is all good for the visual effects industry. There is no
need for data center expansion, which avoids additional

Chapter 5: Looking at Cloud Bursting Use Cases 39
capital investment. When demand spikes, these companies
can do the cost/schedule math to figure out how much addi-
tional processing power is needed, and just rent it from AWS
or Google Cloud Platform.
Because cloud bursting is virtual, the time required to add
additional servers to a render farm to make a release date is
trivial. With no hardware to touch, you can get up and run-
ning with just a few clicks. Just remember the knowledge
Captain Kirk shared with Spock in Star Trek II: The Wrath of
Khan, “Hours instead of days! Now we have minutes instead of
hours!” Cloud bursting makes that possible.
Market Risk Analysis
While developing predictive models based on stock market
data, a leading hedge fund management firm came up with
an idea that could have a significantly positive impact on its
portfolio performance. Thinking the idea had merit, the firm
needed to prove its theory using real data. Getting the data
wasn’t a problem — figuring out how to process it quickly and
cost‐effectively was.
The firm’s quantitative analysts use the Monte Carlo method
to prove out their financial models. Understanding that Monte
Carlo simulations rely on repeated random sampling to get
results and are incredibly compute‐intensive, the firm had to
figure out how to make that happen. The burning question
was how to access the compute power to see if the analysts’
theories were worth changing the firm’s investment approach.
Enter cloud bursting. The firm fired up 30,000 cores using the
Google Cloud Platform to perform its risk analysis. To mini-
mize latency and let those cores access the required data,
the firm used an eight‐node cluster to give those cores the
required storage performance.
Think about that for a second — 30,000 cores! Imagine what
that would have taken ten years ago. The firm would have
placed an order with a server manufacturer. The servers
would have been built and shipped to a data center. Because
30,000 servers generate lots of heat, cooling upgrades would
have happened. At the data center, the servers would have
been uncrated, racked, plugged into power, and hooked up to

the network. After all that work, the application would have
been installed, and finally, simulations could begin.
With cloud bursting, all of that work doesn’t have to happen.
By using the public cloud and providing access to data with
a virtual filer cluster deployed in the cloud, the firm ran
simulations quickly and cost‐effectively, only paying for the
compute it actually used. Even better, the simulation models
could be updated directly by the firm’s quantitative analysts
because of the cluster’s global namespace capability. All of
this resulted in a significant improvement in the performance
of its investments.
What does the future hold? For this particular firm, more
intense simulations, run more frequently. How intense? The
firm has plans to move from 30,000 to 100,000 cores. 100,000!
That’s a pretty serious number. And the firm is confident that
the cloud will be ready with the compute power needed, and
that the storage cluster can provide the performance required
by scaling horizontally.
Genomic Sequencing
H3 Biomedicine (H3) is a Cambridge, Massachusetts‐based
biopharmaceutical company with a bold vision. H3 is com-
bining cancer genomics, synthetic organic chemistry, and
tumor biology to create cancer treatments that are unique to
genetically‐defined patient groups. With a mission to take the
fight to cancer at an individual level, the genomics analysis
processing requirements are immense.
Founded in 2011, H3 is very much a startup. As such, it has no
tolerance for allocating assets that don’t directly contribute
to advancing H3’s core mission. With scientific research at its
core, its computational biologists need to use compute power
the same way its cell biologists use microscopes. Facing nearly
30 percent year‐over‐year data growth, continuing to invest in
local infrastructure is simply impractical. To solve scaling chal-
lenges while giving bioinformaticians access to the tools on
which they rely, H3 turned to the public cloud for horsepower,
using a hybrid cloud approach to ensure storage performance.
Instead of building a massive compute farm, H3 uses AWS
Elastic Compute Cloud (EC2). Just like rendering for movies,

Chapter 5: Looking at Cloud Bursting Use Cases 41
the nature of H3’s workload requires lots of processing power
for intense periods of time. This is a perfect match for the
cloud because many of the jobs scientists run are easily
parallelized.
One advantage H3 gets from a hybrid cloud design is a greatly
simplified architecture and associated operational proce-
dures. Instead of dealing with EC2 block storage, H3’s EC2 and
local compute instances simply interact with the Avere Global
Namespace presented by its storage cluster. This avoids the
tedious process of designing a data synchronization engine to
shuttle data back and forth between local storage and AWS.
Ultimately, this allows H3 to focus its assets on what differen-
tiates its business the most: scientific research.
H3’s on‐premises compute and storage will go through a natu-
ral lifecycle. When these components reach end‐of‐life, the
intention is to move their center of gravity to the cloud. With
a virtual filer cluster as a component of its hybrid cloud archi-
tecture, H3’s Linux tools and applications can talk NFS instead
of being rewritten to use object storage. Financially, using a
virtual FXT cluster to handle the applications’ performance
lets S3 handle the bulk storage without managing AWS’ Elastic
Block Storage (EBS), saving even more money.
With the ability to cloud burst, the vast EC2 compute
resources are able to access the data needed to join the fight
against cancer.
The Sky is the Limit
Five years ago, leading technology companies were just start-
ing to experiment with cloud bursting. Since then, the overall
trend has been towards rapid adoption across industries.
From technology upstarts like Netflix to long‐established com-
panies like General Electric, firms are eager to take advantage
of the benefits the public cloud has to offer.
Technology cycles continue to get shorter and shorter.
Meanwhile, the amount of data in the world is exploding.
Sensor use will continue to expand as smart houses, self‐
driving cars, and smart roads come online. The big data trend
will grow and grow, and humans will continue to search for
new meaning amid this vast data sea. In the search for signal

in all of this noise, demands for compute power will continue
to rise.
As technology improves, the barriers to computing access will
continue to tumble. Companies will be able to use compute
power as they use electricity, confident that the resources
will be there when needed. With tools improving every day,
companies will benefit from inter‐cloud mobility, moving
workloads between cloud providers based on whatever is
cheapest. While cloud computing is becoming more and more
commonplace, sky computing is just around the corner.
You can have peace of mind knowing that a hybrid cloud
approach will ensure that you can burst at will. Your compute
machines will retain high‐performance access to your data,
regardless of which cloud vendor you choose or where your
data lives.

TenBenefitsofHybrid
CloudNAS
In This Chapter
▶▶ Exploring ways in which the Avere Hybrid Cloud can rock your world
This chapter highlights how the Avere Hybrid Cloud NAS
solution specifically helps to tackle both the daily prob-
lems with growing and managing storage and long‐term issues
of trying to keep up with explosive data growth and costs in
your organization.
✓✓ Improving business agility: Whether you are making a
movie, doing drug research, investing money, or develop-
ing widgets, cloud bursting with an Avere Virtual Edge
filer enables harnessing the vast resources of the com-
pute cloud and guarantees on‐time or even early delivery.
This shortens cycle times and lets you get your products
to market faster.
✓✓ Reducing architectural complexity: The Avere built‐for‐
cloud file system enables deploying NAS infrastructures,
even for primary applications, fully in the cloud. Stop
reinventing the wheel; outsource the data center to the
cloud service providers and focus on your business. This
means fewer headaches for you, as well as lower cost.
✓✓ Getting performance where you need it: With Avere,
both physical FXT and virtual FXT Edge filers track usage
patterns of your files and move the most active files
closer to the clients using them. This decreases latency
to clients and reduces the burden on your NAS and
object‐based core filers, improving storage performance.
Chapter 6

✓✓ Keeping pace with growing demand: Demand for more
application performance is growing every year, and this
places an increasing demand on your storage. Avere Edge
filers support clustering from 3 to 50 nodes to help you
get there.
✓✓ Storing data where it works best and costs the least:
Because the Edge‐Core architecture of the Avere Hybrid
Cloud NAS is aware of multiple types of storage media,
each with its own performance and cost characteristics,
your data will always be placed where it makes the most
sense, technologically, and economically — whether
that’s on premises or in the cloud. This lets you swing
your architectural center of gravity to the cloud if and
when it makes sense for your company.
✓✓ Making sure that your data is always available: Avere
Edge clusters support high availability (HA) software
that ensures that your data is always available even in
the presence of network, hardware, and other failures.
✓✓ Managing your data as a single pool: Storage silos be
gone! Avere Global Namespace integrates all your core
filers, both NAS and cloud, into a single pool of storage
with transparent migration between them.
✓✓ Protecting your investment: Hybrid Cloud NAS aug-
ments and extends the capabilities of your existing NAS
environments so that you’re not replacing any gear
you’ve already purchased or are still paying for. There’s
no hurry, so go ahead, follow your equipment’s lifecycle.
✓✓ Consolidating and centralizing management: By gath-
ering together existing NAS environments under the
umbrella of Avere Edge filers, you get a consolidated
view of the performance of your core filers and awesome
storage analytics. This lets you manage storage through
a single pane of glass, instead of several individual man-
agement consoles.
✓✓ Understanding usage and data flow better: The ability
to view what data is being accessed and with what regu-
larity greatly increases your understanding of your active
and inactive storage. This information helps you plan
future purchases with insight.

WILEY END USER LICENSE AGREEMENT
Go to www.wiley.com/go/eula to access Wiley’s ebook EULA.

Cloud Bursting For Dummies Guide

Recommended

Recommended

More Related Content

Similar to Cloud Bursting For Dummies Guide

Similar to Cloud Bursting For Dummies Guide (20)

More from Liberteks

More from Liberteks (20)

Recently uploaded

Recently uploaded (20)

Cloud Bursting For Dummies Guide