Future Of: Cloud Computing

Future of
cloud
computing
A view of the future of cloud computing,
through the eyes of the luminaries who
helped build it.
Table of contents
A stake in the future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Part 1: The enterprise of the future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
How cloud computing will change the way we work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Part 2: Fundamentals of the cloud enterprise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Future of cloud computing

Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Leadership & culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Part 3: What are the impacts for business? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Speed & agility at scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Real-time insights from streaming data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Security & trust . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

Part 4: How should leaders prepare? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Step 1: Find the cloud model that works for you . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
Step 2: Build on an open platform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Step 3: Rethink IT security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Step 4: Lead the change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

A stake in the future
Here at Google, we spend a lot of time looking at data and
anticipating what will happen next. When it comes to the future
of business, we get some of our best ideas by looking at new
information technologies and thinking about their effects. Cloud
computing has long powered Google and is increasingly an integral
part of most organizations’ technology stacks.
But the cloud is much more than a new generation of machines and software.
We’re already seeing that transformation happen to the billions of users and
millions of companies that rely on our technology.
What makes us think that cloud computing will change things beyond IT
Urs Hölzle
itself? History argues for it. Mainframe computers made operations research
possible. The PC revolution made anyone a potential entrepreneur. Client-server
SVP Technical Infrastructure,
enabled corporate reengineering, and smartphones created the app economy
Google Cloud
and computing from almost anywhere.
Future of cloud computing | 5

The cloud, we believe, draws on more diverse data sources, breaks down more
silos, utilizes faster network speeds, and enables more powerful analyses, all
“The cloud, we believe, draws on
at a lower cost than anything tech has seen before. It is not a technology that
more diverse data sources, breaks
belongs to any one provider, or even a bunch of big tech companies. It is a new
generation of computing, a new way of using resources, like mainframes, PCs, down more silos, utilizes faster
and servers were in their time. network speeds, and enables
more powerful analyses, all at a
It is a way of thinking about and doing computing that creates something
lower cost than anything tech has
new: a future with new kinds of product creation, new partnerships, and
seen before.”
new ways of working. This doesn’t mean everything has to change – but
organizations get more from their data, their competitive advantage, and their
existing relationships. – Urs Hölzle, SVP Technical
Infrastructure, Google Cloud
As with any momentous change, business leaders must navigate the here
and now – whether on-prem, hybrid, or multi-cloud – with an eye toward the
future. Leading companies must strive for an even more elevated view as
they embark on their business transformation journey, bringing with them the
experience, history, and strengths of their established systems.
Google’s mission is to organize the world’s information and make it universally

accessible and useful. The company was founded back in 1998, before
anyone could have predicted how fundamental the changes would be in
how humans access and consume information. Along the way, we’ve helped

shape the evolution of how cloud computing works with customers. Our
advances in areas like analytics and machine learning power data collection
“We gathered insights from users,
and productivity, and our security innovations protect billions of users’ data.
conversations with businesses
It’s been 20 years now, but these changes are far from over. In fact, for many
companies, they’re just beginning. of all sizes, surveys, polls, and
knowledge from within Google
This is why we assembled this report – to help business and IT leaders itself to offer you a holistic view of
navigate the changing business world with cloud computing. We gathered
the current landscape and where
insights from users, conversations with businesses of all sizes, surveys, polls,
it’s already going in terms of value,
and knowledge from within Google itself to offer you a holistic view of the
current landscape and where it’s already going in terms of value, innovation, innovation, and competition.”
and competition.
– Urs Hölzle, SVP Technical
The goal is to give you the insights you need to craft winning strategies by
building on your existing environment and moving toward a future that unlocks
the benefits of cloud computing. We hope that you can put these examples
from real companies, current market trends, and learnings from Google
executives to good use, and as always, we look forward to partnering with you
on whatever comes next.
– Urs Hölzle, SVP Technical Infrastructure, Google Cloud

PART 1
The
enterprise
of the
future
What makes new computing technologies great? It’s not just
networking speeds, compute power, or storage capacity. Those [Our customers] are gaining
are critical minimums, and cloud computing can deliver them far something new in cloud computing:
more effectively than previous methods. There is more, however. speed, efficiency, reliability, and
The real value of technology is not what it’s made of, but what security in service of a new wave of
it does.
business-driven innovation.
Great IT enables a company to do the things it’s best at, by getting low-
value work out of the way. That means more time to focus on delighting
customers, the core of any successful business. Innovation fuses hard-won
knowledge about customers, products, and markets with new capabilities,
enabled by transformative IT.
This is why so much of the economy runs on and was shaped by evolving
technologies. For a reasonable investment, technology has provided benefits
like empowerment, improved teamwork, and services on demand. In fact,
some technologies – like cloud computing – offer all of this at a lower
overall cost than their traditional alternatives.
A decade since businesses first started to run on public clouds, many have
discovered that cost savings were just the beginning. Our customers are
creating businesses that are continuous and nearly frictionless, maximizing
their organizations’ missions and leveraging current resources with

faster times to insight and action. They are gaining something new in cloud
computing: speed, efficiency, reliability, and security in service of a new wave of
business-driven innovation.
If the value of a technology is the net experience it delivers to the enterprise

– on cost, on performance, and on time to innovation – it’s clear why so
many believe that cloud computing’s interconnected, continuous streaming
will be an important part of our future economy. In other words, the future of
business will depend on realizing the promise of cloud computing.
In the following pages, we’ll explore:
➊ What the enterprise of the future will look like
➋ What fundamentals shape the cloud enterprise
➌ How cloud computing will impact businesses
➍ How business and IT leaders and decision makers can prepare

From the data center, to the cloud, The growing cloud
to the edge
By 2024, most enterprises
will have intensively multi-
90%+ cloud environments, with
Cloud computing is already an integral part of most organizations. According on-prem, off-prem, public,
to a survey by RightScale,1 81 percent of companies with 1,000 employees and private cloud.6
or more have a multi-platform strategy. By 2024, that number is expected to
climb to more than 90 percent.2 Between 2018 and 2021, worldwide spending
on public cloud services is expected to grow 73 percent from $160 billion to
$277 billion.3
2018 $160 billion
Companies are moving to the cloud because it offers the agility they need
to compete in a fast-moving, rapidly changing customer environment. Cloud
saves companies the cost of maintaining their own internal systems; it drives 2020 $277 billion
innovation by making it easier to develop and change applications on the fly;
and it helps standardize and simplify security management.
Global spending for public cloud services is
The next few years, according to data from analyst firms, will see cloud on track to reach $277 billion in 2020.7
computing continue to evolve as part of a stack that includes the Internet
of Things (IoT) and edge computing, or processing at the source of data
ingestion. Gartner predicts that by 2025, 75 percent of enterprise-generated
data is created and processed outside a traditional centralized data center or
cloud.4 In addition, more than 40 percent of organizations’ cloud deployment
will include edge computing, and 25 percent of endpoint devices will execute
AI algorithms by 2022, according to IDC.5

Businesses are quickly adopting
How cloud computing will change cloud computing
the way we work 77%

At its most optimal, cloud computing enables businesses to focus on
performance as directly as possible, with minimal hassle or interruption. New of decision makers* say they will use cloud for
advantages built upon the foundation of cloud computing – agility, mobility, a majority of their needs by 2029.8
flexibility, and security – emerge when highly scalable and flexible computing
technology fuses with existing technology and processes (whether on-
premises or in the cloud). The aim is to get the computation you want and A cloud for everyone
need, the way you want it, and move on.
64%
Let’s begin with some examples from enterprise technology.
of decision makers* believe everyone will
In software development, these include the continuous creativity of agile have access to computing by 2029, including
programming and the “launch and learn” characteristics of continuous remote regions.9
integration and continuous delivery (CI/CD), along with the explosion of open-
source software (a faster and more responsive version of the standards
bodies and years-apart release cycles that used to be industry norms). Cloud: the new growth lever
In technology security, elements like BeyondCorp, Google’s zero-trust enterprise

87%
security model, envision work happening from anywhere, from any kind
of approved device, replacing the previous standard of firewalls and other
of decision makers* expect cloud computing
boundary-oriented practices. Devices stream their states to a managing host,
to become an important driver of revenue
and are thus permitted access. Access is managed on the user and device
growth by 2029.10
level, rather than the network level.
*IT and business decision makers at global

mid-market and large companies

In data collection and analysis, streaming data now feeds into existing
machine learning systems for continuous learning and adjustment (see
The promise of cloud computing represents
sidebar, “Streaming insights to feed the world”).
more than just a faster, more efficient version
Moving into this new world means transitioning to a new way of working that of everything we did before – it’s an entirely
assumes continuous flow. Products are never “finished”; there is constant new way of thinking and working across any
updating and iteration. Security is not a linear process based on static events, IT environment.
but rather an ongoing practice. Data analysis isn’t just faster; it’s continuous
and iterative. The promise of cloud computing represents more than just a
faster, more efficient version of everything we did before – it’s an entirely new
way of thinking and working across any IT environment.
It stands to reason, then, that this creates a social impact, both internally and
with customers, enabled by technology.
➊ Collaborative technologies like document sharing end the stop/

start practice of versions, in favor of a continual process of creating,
editing, and commenting. Cheap and easy video at the click of a button
strengthens communication, speeding actions.
➋ Prototyping and personalization are faster and easier, thanks to rich data
streams in both directions.

➌ In product lifecycles, customer relevance is increased by cloud-
connected objects subject to frequent software adjustments. This
might be new versions of subscription-based software, or even
alterations in the performance of hardware itself (like in 2018 when
Tesla improved the braking capability of its cars with an over-the-air
download). As awareness of customer behaviors increases, customers
consciously and unconsciously raise the bar, effectively demanding
that they be understood and their next need anticipated.
➍ Even slow-changing aspects of business, like supply chain and

logistics, are moving toward streaming capabilities. Blockchains
enable real-time spec changes and payments. Virtual warehousing
and Uber-type delivery services speed up slower processes by making
them continuous. Cloud-based companies like Shippabo optimize
route management and automate compliance for faster action.
There are many more examples. To fully understand how to prepare for the
future, however, it’s helpful to first examine how these attributes emerge
within the enterprise.

A father of the Internet looks to the future
It’s no exaggeration to call Vint Cerf a father of the Internet. Together with Bob
Kahn, he developed the TCP/IP protocols that are the basis of the Internet.
Now, as Google’s Chief Internet Evangelist, Cerf travels the world talking and
writing about what comes next.
“When we were designing the Internet, one of the things that we didn’t care about
was the details of each network, because we imagined hundreds of thousands
of networks all interconnected to each other,” Cerf recalls. “That’s where cloud
computing came from: The whole idea was that the computers you were using
were somewhere in the cloud, connected to the Internet somewhere. You didn’t
care where.” The division between public clouds is a temporary one, in Cerf’s
view. “There’s now a recognition that the interaction between the clouds and the Vinton G. Cerf
movement of data back and forth between the clouds makes sense,” he says.
VP and Chief Internet Evangelist,
Cerf believes we’re entering the next golden age of computing, where artificial Google
intelligence and edge technology will bring computing to every corner of our
lives until it’s pervasive enough to go unnoticed.
“More and more computing and software are closer and closer to us, to the
point where we’re wearing it or it’s embedded in us,” he explains. “The distinction
between the online world and the offline world starts to evaporate because
the online world has penetrated so deeply into physical spaces everywhere. At
some point, it could be that this kind of communication and computation will be
as ubiquitous as we hope electricity is.”

This, of course, brings risk. “We are building yet another dependency into
our social and economic fabric, and we should be giving that serious
“I think that software producers, individuals
thought,” Cerf says. “I want to urge people to think more deeply about the
and corporations should feel an increased
kinds of infrastructure that we are dependent upon and ask: How do we
make it resilient?” sense of ethical responsibility for assuring
that whatever their products are, they don’t
He’s encouraged by conversations about the ethical implications of software turn out to be harmful, even by accident.”
development – not just about building it ethically, but also being aware
of potential vulnerabilities. “I think finally, questions are being asked. I
– Vinton G. Cerf, VP and Chief Internet
think that software producers, individuals, and corporations should feel an
increased sense of ethical responsibility for assuring that whatever their Evangelist, Google
products are, they don’t turn out to be harmful, even by accident.”
In Cerf’s view, better products stem from a clearer understanding of how

those products work. This could mean including software or security
engineers in product development meetings from the beginning so that the
product strategy incorporates technical infrastructure, data management,
and security considerations. “It’s really a new model of the role of technology
inside the business,” Cerf says, “where the developer is part and parcel of the
product offering.”
But Cerf is also a big thinker, and he doesn’t hesitate to explore the topic
of an interplanetary version of the Internet, designed to connect humans
and machines as we venture beyond our galaxy. “Certainly by the end of
the century, and maybe before that, we will have a stable interplanetary
backbone to support our exploration into the solar system,” he says, an
explorer’s twinkle in his eye. “The universe is full of problems, and I’m an
engineer. I like solving problems.”

Looking to the cloud for ERP Solutions
When it comes to investing in enterprise software, most companies are not

dazzled by gee-whiz technology. Instead, they have a singular demand: a
platform that can provide the data and analysis necessary to gauge what the
marketplace demands and deliver value. “It’s what every enterprise has been
after for the last 20 to 30 years,” says Pavan Srivastava, a principal at Deloitte
Consulting LLP, who leads the SAP on cloud practice. “But the technology just
wasn’t there yet at a cost attractive for mass adoption.”
This is hardly the case going forward, and Pavan is discovering signals of this
change every day. When SAP companies describe what they want from their
ERP, he almost always finds a way to deliver in the cloud. That’s partly because
the cloud platform centralizes mass amounts of decision-making data from
sources with speed and performance previously unimaginable. Retailers, for
instance, can now install sensors in their shelves that gather and analyze data
to help them understand what products consumers want most, and when to
replenish those popular items to bolster sales and reduce waste.
Better yet, cloud-based tools can deliver real-time analysis in a steady stream,
transforming otherwise useless bits of data into actionable items. Pavan
points to a dairy manufacturer who measures the temperature and weight
of cheese and butter flowing through their production line with sensors.

By sending this data to the cloud where machine learning tools study its
patterns, manufacturers can determine what portion of their products have
“The innovation of the cloud comes from
already spoiled before shipping and devise a strategy to reduce the number of
pooling and sharing these kinds of capabilities
bad cartons arriving in stores.
so that companies can consume tools, like
Cloud computing also removes the burden from companies to process data machine learning, as a service rather than by
themselves. Semiconductor manufacturers typically run simulation tests on building it themselves.”
their microchips once or twice a month – a task that requires quickly spinning
up hundreds of virtual machines for a short period of time and costs millions of
– Pavan Srivastava, Principal,
dollars in hardware to deploy. By conducting these tests in the cloud, businesses
can reduce their expenditure to a pay-as-you-go subscription model rather than Deloitte Consulting
wasting money on limited-use technology that lays dormant most of the year.
With these newfound savings, they can run tests more frequently, keeping regular
tabs on their chips’ performance and making improvements more quickly.
The cloud can fuse these solutions together. As Pavan says, “The innovation of
the cloud comes from pooling and sharing these kinds of capabilities so that
companies can consume tools like machine learning as a service rather than by
building it themselves.” For enterprises, that translates into more time spent on
innovating and delivering value to customers, a boon to everyone involved.

PART 2
Fundamentals
of the cloud
enterprise
The qualities of cloud computing – speed, flexibility, and continuity – Edge computing goes mainstream11
manifest in different ways within different companies, and even vary
within cloud stacks. Large-scale clouds, of course, play a key role in
the organization, analysis, and management of much of this. Equally
important is the vast increase in the number of edge devices built to ●●●
send and receive new information, whether in the form of connected
1 in 3 companies* use edge computing for a
computers, IoT sensors, or purpose-built appliances.
majority of their cloud operations.
An increasingly important dimension is the advent of cloud-type technologies

inside and alongside existing technology. Kubernetes, the highly popular
container management platform, and Istio, a microservice management
layer, were both initially open-sourced by Google to extend the benefits 66%
of cloud computing to the rest of the IT stack. As with any open-source
more companies* will use edge computing for
solution, adopting tools like these can spur new levels of information-sharing
a majority of their cloud operations by 2029.
and interoperability with existing public and even private clouds, further
strengthening the trend toward streaming.
In this section, we’ve taken a technology-driven approach to analyzing how

the enterprise of the future takes shape – from the specific tools a company
of companies* believe
adopts, to the internal systems these tools create, to the company’s evolving
that by 2029, edge
culture and leadership. Although rooted in technology, each of the examples 83% computing or IoT will
we explore holds broader implications for the rest of the business.
impact every industry.
*Global mid-market and large companies were surveyed.

Tools
For many companies, cloud adoption starts with a simple objective: handling
core computing tasks like storage, networking, and application management
With storage that expands to need, and the
more cheaply and efficiently. And of course, everyone’s universe of data is
growing. According to IDC, the total volume of global data will reach 175
networking and computing power that can
zettabytes by 2025.12 To stay relevant, companies need to manage and make bring the Internet places where it has never
use of that. The cost of storing data on-premises starts with the investment been before, it’s no wonder so many companies
in physical storage devices, but new costs appear in the form of storing, are moving their data to the cloud.
updating, and maintaining these machines.
It’s not just a volume problem. The variety of data formats and sources (IoT
devices, online customer interactions, legacy business systems, and so on),
as well as the velocity with which data is created, processed, and analyzed,
create additional complexity. Getting all of these to play well together requires
a network that is secure, reliable, and fast enough to process them all in real
time, even in areas with low accessibility and high latency.
To meet these challenges, companies are turning to the cloud. By moving

data out of physical servers, businesses can take advantage of virtually
unlimited storage while saving money on upkeep. These companies also
unlock cloud computing’s other networking and computing benefits, such as
high reliability, minimal downtime, and complementary tools to facilitate data
processing and analysis.
With storage that expands to need, and the networking and computing
power that can bring the Internet places where it has never been before,
it’s no wonder so many companies are moving their data to the cloud.

In fact, according to research by Cisco, public cloud When applied in the right use case, the business
workloads and compute instances are going to grow impact of streaming analytics can be significant:
by 28 percent from 2016 to 2021.13 Google itself has used streaming analytics to cool
its data storage centers, generating up to 40 percent
Once the data is secured in the cloud, it can easily be in savings (see sidebar, “How Google reduced data-
processed and managed to power a diverse set of center cooling costs”). Businesses don’t need to be
advanced tools to provide greater speed, flexibility, fully on the cloud to benefit from streaming analytic
and functionality to application development, IT services, Hölzle notes. For example, data analysts
operations, data analysis, and more. According to Urs can send a copy of on-premises data to the cloud
Hölzle, one of the company’s original employees and and receive real-time analysis within seconds.
the SVP of Technical Infrastructure for Google Cloud,
“Google initially built a high-velocity information To further extend the value of their data, many
environment – a platform where massive amounts of companies are beginning to rely on machine learning
data can be transmitted and analyzed at a rapid pace (see sidebar, “Cloud computing unlocks the AI
― to support its search engine. Now that the cloud potential”). After decades of stop-start progress, ML
can expand storage of its data delivery, the demand is now a reality due to the recent convergence of
for high-velocity is universal – and the pace and scale three factors: advanced algorithms, access to huge
will only increase.” volumes of digital training data, and cheap, abundant
computing power – available, of course, through
At Google, streaming analytics – the ability to the cloud. ML tools can help businesses identify
process, analyze, and act upon data in real time – is patterns within huge, streaming datasets; automate
“actually the fastest-growing analytics sub-case that repetitive tasks; and free humans to spend more time
we have” and will continue to grow, Hölzle says. on creative, non-automatable work (like writing new
applications or crafting great user experiences).
How Google reduced data-center cooling costs by up to 40% using AI
Reducing energy consumption has long been Using the historical data collected by thousands actions from the PUE model, to ensure that we
a priority for Google. We have built our own of sensors within the data center – including do not go beyond any operating constraints.
super-efficient servers, invented better ways temperatures, power, pump speeds, setpoints, We then tested the model by deploying on a live
to cool our data centers, and invested heavily and more – we trained an ensemble of deep data center.
in green energy sources, with the goal of being neural networks on the average future PUE
powered 100 percent by renewable energy. (power usage effectiveness), defined as the The result? Our machine learning system
One achievement we’re particularly proud of is ratio of the total building energy usage to the IT was able to consistently achieve a 40 percent
using machine learning to reduce the amount energy usage. We then trained two additional reduction in the amount of energy used for
of energy we use for cooling by up to 40 ensembles of deep neural networks to predict cooling, which equates to a 15 percent reduction
percent. This not only helps us reach our energy the future temperature and pressure of the data in overall PUE overhead after accounting
efficiency goals, but also passes energy savings center over the next hour. The purpose of these for electrical losses and other non-cooling
on to our customers. predictions is to simulate the recommended inefficiencies. It also produced the lowest PUE
the site had ever seen. Because the algorithm
is a general-purpose framework to understand
complex dynamics, these results can be applied
in a range of other scenarios.
Learn more about our data center approach at

google.com/about/datacenters
Source: DeepMind blog

Cloud computing unlocks AI’s potential
Thanks to the cloud’s ability to deliver an helping law enforcement officers apprehend child
unprecedented level of computing power at abusers. They can help analysts spot patterns in “Machine learning is a
relatively low cost, artificial intelligence – broadly unstructured data such as images and videos,
natural outgrowth of
defined as the ability of machines to exhibit automate data analysis for faster insights, and
cloud computing.”
intelligent behaviors such as learning, complex merge historical data with real-time inputs to
problem-solving, and natural language recognition anticipate security vulnerabilities or product
– and its subset of machine learning are becoming failures. But for many companies, bridging the gap – Urs Hölzle,
mainstream. A survey by MIT Technology between possibility and reality is the tricky part. SVP Technical
Review shows that, in the business world, AI Infrastructure,
and machine learning are happening now.14 The “For customers to realize AI’s full potential, cloud-
Google Cloud
majority of respondents (60 percent) have already based ML tools must be accessible enough for non-
implemented machine learning strategies, and experts to use without any IT assistance,” Hölzle
nearly one-third considered themselves to be at a explains. “While this is the case in some areas of
mature stage with their initiatives. cloud computing, we are not there yet.” But cloud
providers are hard at work on solutions, and tools
“Machine learning is a natural outgrowth of cloud like Google Cloud AutoML and machine-learning
computing,” says Hölzle, Google Cloud’s SVP of APIs offer a starting point to an exciting future.
Technical Infrastructure. “If there are massive
amounts of data stored on the cloud, then the “The beauty of ML is that it allows analysts to look
obvious next step is to use built-in tools to make at data in several different iterations, and ask new
sense of the data and predict future outcomes.” questions with each one – thus developing a much
deeper understanding of their data,” Hölzle says.
AI and ML have broad implications for business – “That quality will expand as ML tools on the cloud
from helping farmers sort cucumbers in Japan, to become more accessible.”
improving the accuracy of medical diagnoses, to

Systems
Cloud-based tools often represent the first their own special way, because there’s been no most of the audit is automated and the stack
step in a company’s digital transformation, and notion of ‘IT as a service’ for the last 10 years.” is already audited,” he says. By contrast, every
frequently lead to more systemic changes. When Cloud platforms are helping to standardize a proprietary on-premises environment is unique,
applied systemically, cloud computing can usher company’s approach to IT infrastructure and so automation isn’t always possible (or scalable).
in a new IT model centered on standardization, management, which in turn can eliminate “There are too many cases, and the test matrix is
interoperability, and openness. This shift can have some of the “mundane work” of creating and too big.”
powerful business implications, such as reducing re-creating custom integrations. And the less
operating costs, increasing efficiency, and even time a company’s IT team spends on mundane, Cloud computing holds similar promise for
making customers and employees happier. Let’s undifferentiated work, the more time and energy security. “Imagine you have a million machines,
take a look at how these systems evolve. it can devote to creating differentiated value that and they’re all reporting on their security status,”
supports the company’s mission. says Heather Adkins, Director of Information
Standardization Security and Privacy at Google. “You could look
Standardizing on cloud platforms also creates across the baseline of all these machines at the
Over the past 20 years, enterprise IT has spent new opportunities for automation at scale, which same time and pick out the anomalies. That’s
considerable energy integrating disparate off- can help companies more effectively manage the power of operating in a centralized model.
the-shelf solutions to create a system that works compliance and security. The old IT way of thinking was to buy a bunch of
for its unique needs, says Google Cloud CTO machines and plug them all into a network. But
Brian Stevens, an industry veteran who spent In a few years, Urs Hölzle envisions, the cloud that means you also can’t measure whether your
nearly 13 years as CTO of Red Hat before joining will make compliance much easier. “For security is good.”
Google in 2014. “They all build off common sets example, it should be possible for you to get a
of products, but they each knit it together in HIPAA audit for a few hundred dollars because

Interoperability & openness
Open-source software plays a key role in while ensuring these tools and systems work
standardization across tools, systems, and clouds together allows them to maintain speed and scale
(see sidebar, “Bridging multi-cloud and on-premises throughout the stack. As more companies adopt
environments”). By definition, vendor-neutral, open- hybrid and multi-cloud strategies (more than 90
source software is designed to work with any percent will by 2024, according to IDC16), open source
platform (and any functionality that doesn’t exist will only grow in importance.
from the start can be built and shared by an active
community of third-party developers). This protects Openness and interoperability do not mean more
companies from betting too heavily on a single complexity or a loss of control. On the contrary,
vendor and then having to replace the entire platform companies can leverage the benefits of the cloud
if or when it becomes obsolete. Adoption trends bear while continuing to meet business needs. They get
out the value of open-source solutions. According to central management, a more seamless transition to
a 2018 survey by The New Stack, more than half of the cloud, clear choice, and protection from vendor
companies are likely to run open-source software, lock-in. Open-source solutions, like containers, are
and when zooming in on enterprises with more than foundational tools that bridge multiple cloud and
1,000 employees, the adoption rate of open-source on-premises environments to allow for consistent
software jumps to 63 percent.15 management across any environment. With a hybrid
cloud environment, businesses can “modernize
In addition to incorporating open-source solutions in place” or “move and improve” applications to
into their IT strategies, streaming enterprises also the cloud to reap benefits with less effort. Open
prioritize interoperability. Choosing from a mix of architectures give IT leaders the flexibility to tap the
proprietary, open-source, and even home-grown tools best solutions that meet their unique business needs.
and platforms gives companies maximum flexibility,

Bridging multi-cloud and on-premises environments
For most organizations, moving to the Start by leveraging your existing resources. Containers to future-proof businesses17
cloud is a gradual process. The longer a Whether you’re in an on-premises, hybrid, or
company has been in business, the more multi-cloud environment, you can start by
it’s likely to have invested in legacy data and modernizing your applications in place to keep
IT systems, along with the skill, talent, and
institutional knowledge to manage those
your existing systems while incorporating
innovative cloud services on your own terms.
●●●
systems. Transitioning to a cloud platform 1 in 3 workloads are containerized across
can be difficult and costly for businesses In Stevens’ view, stay focused on on-prem and cloud in the enterprise segment.
that have to manage this change while also standardization, openness, and interoperability.
keeping old systems and teams intact. In Tools like the open-source container
heavily regulated industries such as financial orchestration platform Kubernetes exist to
help IT teams build, deploy, and manage
services and healthcare, certain workloads 46%
may need to remain on-premises indefinitely, applications consistently across multiple cloud
making hybrid IT a long-term – rather than and on-prem environments. Containerized workloads are expected to
temporary – strategy. grow by 46% in the public cloud by 2021.
But there’s also the challenge of keeping
As Google Cloud CTO Brian Stevens puts it, IT existing talent engaged while reallocating
teams “are living in this world of pain, of legacy certain resources to new projects and hybrid
systems and architectures that, to be honest, integrations. Stevens recommends starting
is going to be there for the next number of with small teams and small projects that
years – and meanwhile they are using public can yield quick wins in a matter of weeks,
cloud technologies for many of their newer then scaling gradually as more workloads
workloads and data analytics projects.” are deployed on cloud platforms, in hybrid-
friendly formats.
So what’s a leader to do?

Leadership & culture
Even with the right tools and systems in place, success in the cloud can’t
happen without the leadership to guide it and the culture to achieve it. And,
as a company transitions to its streaming future, new tools and systems in
turn affect company culture.
“[Migration to the cloud] doesn’t necessarily start by changing the whole

organization at once,” Stevens says. “It just starts by getting a small group
of people really subscribed to the mission, and then they bring everybody
else along.”
For example, a company might choose to containerize applications so

they work across multiple cloud and on-premises environments. It might
choose an open-source container management layer like Kubernetes
to further support its multi-cloud strategy and need for interoperability
among different environments. The integration of an open-source solution
into its IT stack might increase enterprise developer participation in the
Kubernetes community, which might lead to a cultural shift to prioritize
openness and transparency. This, in turn, might lead to the adoption of
workflows that support these values, like a development and operations
(DevOps) model in which both software developers and operators are
jointly responsible for outcomes and openly share feedback in blameless
postmortems. Ultimately, the cultural shifts in this example can yield
significant improvements in productivity, agility, product quality, reliability,
and security. In the next section, we’ll delve into specific examples of what
these improvements look like in the real world.

The future of news
About a decade ago, the news industry was in the midst of In one move, cloud became the default position for most
a major transition as readers were opting for smartphones major technology purchases and implementations. “We have
over paper. While some publications struggled with the discovered ways to use the [cloud] as an agent for change,”
digital migration, U.K.-based Telegraph Media Group saw it says Wright.
as an opportunity to connect with readers on new platforms,
launching its most radical transformation since The Daily From there, Telegraph Media migrated to a public cloud
Telegraph’s inception in 1855 – all while using resources and then on to a multi-cloud platform. The IT team adapted
constricted by the 2008 recession. application programming interfaces (APIs) to distribute news
to all mediums via the cloud. (Last year, the media company
Chief Technology Officer Toby Wright had a bold new made around 460 automated release-updates to its content
digital vision. Properly delivering that vision required talent API.) Journalists started using popular productivity tools
– technology professionals capable of creating apps and that seamlessly integrate with G Suite to collaborate on
websites to resonate with the new generation. And, while complicated stories written by contributors from all over the
Telegraph Media had plenty of smart people in their IT globe. And the technology department built specialized tools
department, the day-to-day maintenance of the existing legacy to speed up certain content creation processes. For example,
infrastructure required much of their attention. compiling a photo gallery to accompany an online story used
to take as long as 18 minutes, an eternity by today’s publishing
That’s when Wright and his team discovered that the standards. A new custom-built, cloud-based authoring tool
resources his team required were in the cloud. They started can do the job in just a few minutes. In addition, Telegraph
small by boosting collaboration with G Suite. This eliminated Media could process up to 4 TB of analytical data in less
the need for infrastructure maintenance, freeing Telegraph than a minute, which drives the tools they use to provide
Media’s IT people to focus on product innovation. In one more relevant content. To Wright, this “means as long as
move, the technology team captured IT services and a new we have an idea, we can pretty much do it, and at low risk.
revenue-generating engine with the same budget and staff. Experimentation at scale is much easier.”

This change has also allowed the existing technology team to be a draw for talented IT and technology professionals looking
build their own tools, enabling operations people and software to become fluent in different cloud universes. Multi-cloud also
developers to work seamlessly – the basic tenet of the DevOps creates a great opportunity for vendors to compete on features
model. “Culturally, that leads to a strong sense of accountability and and customer satisfaction instead of locking businesses into
responsibility,” explains Wright, “because engineers cannot, nor will multi-year contractual agreements.
they, just throw software over the wall to an operations team.”
Today, leveraging the cloud is the norm for Telegraph Media ➋ Embrace open source: Open source makes it easier for your
employees. Here are Wright’s top three recommendations to help team to learn how to build software while helping to give them
companies transition smoothly to the cloud: a solid understanding of the technologies that underpin cloud.
Being able to “see” into the black box is very useful.
➊ Choose your platform wisely: Whether you’re thinking

about using single- or multi-cloud services, make sure your
➌ Trust your people. Let employees experiment with cloud-
structure fits your company’s needs. Telegraph Media recently
based tools on their own. Engineers have access to the whole
announced plans to operate mostly on Google Cloud, but Wright
spectrum of Google Cloud technologies and don’t have the
sees advantages to both. Using one cloud guarantees fluidity
hurdle of needing permission to experiment. Journalists at The
and one-stop-shop service, while multi-cloud platforms can
Telegraph uncovered new purposes for Google’s cloud-based
provide rapid access to a sea of best-in-class solutions that fuel
collaboration tools that Wright had never even anticipated.
continuous innovation. In addition, multi-cloud platforms can

PART 3
What are the

impacts for
business?

The enterprise of the future represents the realization of cloud computing’s
original promise: speed and agility at scale, real-time insights from a large volume “The things that we cared about
and variety of data, and improved security and reliability. Each of these changes [20 years ago] are pretty much
also affects the way people work – particularly the people writing and operating the things we care about today:
cloud services. flexibility, cost attractiveness,
Melody Meckfessel, VP of Engineering at Google Cloud, leads Google’s DevOps practice, and
a stack that gets updated every
she’s seen firsthand how culture plays into a company’s transition to the cloud. “What we’ve day. What really has changed is
learned within DevOps specifically is that the productivity, speed, and quality that comes the depth of how all of that has
with using the automation – and the culture within how you develop software – is powerful, played out – through open source,
and it leads to better business outcomes,” Meckfessel says.
through ML stacks, through content
At companies like FTD, a premier floral and gifting company with operations in the U.S. and
delivery networks. The offering
U.K., the impact of cloud technology is already clear (see sidebar, “How the cloud helped today is much more complete than
keep a 110-year-old company relevant”). Recognizing that it needed new tools and systems it was back then, but really, the
to compete in a streaming world, FTD migrated to Google Cloud Platform. This change principles are the same.”
enabled FTD to deploy services 10 times faster and helped foster a culture of continuous
integration and delivery – which ultimately helped ensure that the company was better
– Urs Hölzle, SVP Technical
meeting the needs of a demanding digital customer base.
In this section, we’ll explore how realizing the full potential of cloud computing can impact
businesses across industries and geographies.

How the cloud helped keep a 110-year-old
company relevant
FTD Companies, Inc. (FTD) has provided floral “We couldn’t afford to be held back as we
products and services to consumers for more reinvented the company, and we didn’t want to
than a century. Its iconic “Mercury Man” logo put a Band-Aid on our existing infrastructure,”
is displayed in approximately 35,000 floral explains Vamsi Muddada, Chief Technology
shops in more than 125 countries, reflecting Officer at FTD Companies. “Once we decided to
the widespread popularity of the company’s move to Google Cloud Platform, we never looked
FTD and Interflora brands. Today, FTD operates back. Our lives became so much simpler.”
a diversified portfolio of brands, including
ProFlowers, ProPlants, and Shari’s Berries. FTD’s business is seasonal, with peaks during
the most popular floral and gift-giving holidays
FTD understands that in order to be successful, it such as Valentine’s Day and Mother’s Day. FTD
must focus its efforts on technology and develop needed its new cloud infrastructure to be ready in
innovative new software and services. To move time to take on a portion of the holiday load, so IT
into this new world, it needed to get out of its data enlisted Google Cloud Professional Services for
centers and into the cloud. Based on flexibility, a four-week engagement.
cost, and the variety of managed services
available, FTD decided to build a cloud-native “In just four weeks, Google helped us design an
architecture on Google Cloud Platform to speed ecosystem of Google Cloud Platform services
development for its new ecommerce sites. architected in the most optimal and cost-effective
way,” Muddada says. “Getting to that point on our
own would have taken us a year or more.”

FTD now runs a container-based architecture FTD now follows a continuous integration/
based on Google Kubernetes Engine (GKE), continuous delivery (CI/CD) model, using With its CI/CD pipeline on GCP,
using Terraform by HashiCorp to automate Jenkins to automate portions of the the company expects to be able
cluster deployment and Stackdriver for development process to continuously deliver to make the most of its seasonal
monitoring. Data is stored either in Cloud new features and accelerate app innovation.
sales peaks, with improved
Datastore, a fully managed NoSQL database With its CI/CD pipeline on GCP, the company
service for web and mobile applications, or expects to be able to make the most of conversion rates and related
in Cloud SQL MySQL databases. With GKE, its seasonal sales peaks, with improved revenue benefits.
FTD can deploy services 10 times faster, conversion rates and related revenue benefits.
accelerating time to market for new software
and features. With automated deployment “With Google Cloud Platform, we were able
and scaling, developers were able to ramp up to drive a cultural shift toward continuous
quickly and improve their productivity. integration and DevOps that puts us in a much
better position to compete,” Muddada says.
“GKE is a beautiful thing,” Muddada says. “It’s such a developer-friendly platform that we
“Within a day we can write, build, and spin up were able to get buy-in very quickly. Everyone
a new service that would take us two weeks to loves working on GCP because it’s so easy.”
deploy on-premises and a week to configure
with any other cloud provider – and we don’t Adapted from cloud.google.com/customers/ftd
have to manage the clusters ourselves.”

Speed & agility at scale
In the streaming enterprise, change is a constant, and operating at the speed of digital activity
and data is crucial to delivering on customer expectations. This is true across industries – from
personalization in retail to high-performance computing in finance to supply-chain optimization
and reliability in manufacturing. The streaming enterprise must have the tools, systems, and
culture in place to enable real-time action and constant agility at scale.
Take, for example, human genomics, the science of studying patterns within human DNA.
It increasingly relies on high-performance compute and storage resources.
The Broad Institute of MIT and Harvard, a collaboration of MIT, Harvard University, and Harvard-
affiliated hospitals, studies the human genome to reveal the secrets behind the origins of
diseases and to help find new cures and therapies. The Broad Institute builds on the success of
the Human Genome Project, the international research effort to sequence and map the genetic
instructions inside each of us.
A single human genome contains more than 3 billion base pairs of genetic material. For
accuracy, researchers typically examine each base pair approximately 30 times for sequencing,
meaning they gather almost 100 billion base pairs worth of raw data – nearly 100 gigabytes –
per person.
In 2018, the Broad Institute generated one human genome equivalent every eight minutes –
roughly 16 terabytes of data every day.
That’s where Genomes in the Cloud and Google Cloud Platform come in. Together, they let
Broad Institute researchers continually analyze data from thousands of samples each year
without having to worry about delays or interruptions to their potentially life-saving work.

Moving to Google Cloud meant the Broad Institute could analyze human genomes 400 percent Faster feedback, quicker releases,
faster, supporting researchers in their efforts to find new cures and therapies.
more innovation
“We can do important research faster than ever,” says Geraldine Van der Auwera, associate
director of outreach and communications for the data sciences platform at the Broad Institute.
“That will lead to a greater understanding of the human genome and the links between
60%
genetics and human disease.”
of decision makers expect to update code
For many companies, the way applications are created, updated, and managed is a natural weekly or daily by 2029, up from 37% today.19
starting point. The rise of DevOps – a software development model that combines software
development (dev) with IT operations (ops), now in use at 60 percent of companies globally,
according to Harvard Business Review Analytic Services18 – and growing popularity of CI/
CD indicate that more and more IT organizations are embracing a streaming approach to
application development and IT management (see sidebar, “The state of DevOps”). of respondents say
it’s important to their
Often, increased speed and agility result from improvements (even minor ones) in developer and 86% company to develop
IT operator productivity. “If you go in and look over the shoulder of a developer or an IT person, it and release new
can be eye-opening how much complexity they have to deal with,” says Stevens, Google Cloud’s software quickly.20
CTO. “There’s so much inefficiency that they have to go through just to do the impactful part of
their job.” In one company after another, Stevens has seen how cloud technology can reduce or
eliminate those inefficiencies to yield “massive impact.” By standardizing on cloud platforms,
Stevens says, “you can dramatically change how they spend their day...and if organizations
can redeploy their staff and move away from the cost center that IT is today into a value center,
that’s exciting.”

The state of DevOps
DevOps is rapidly gaining traction among companies as a means of ➌ Open-source software improves performance.
improving speed, efficiency, and quality. In 2018, Google sponsored Open-source software is 1.75 times more likely to be
DevOps Research & Assessment’s (DORA)* annual survey on the extensively used by the highest performers, who are also 1.5
state of DevOps, which polled nearly 1,900 DevOps professionals to times more likely to expand open source usage in the future.
distill the practices that set top-performing DevOps teams apart.
Some key highlights from the findings: ➍ Outsourcing by function can hurt performance.
While outsourcing can save money and provide a flexible
labor pool, low-performing teams are almost four times
➊ Availability matters.
more likely to outsource whole functions such as testing or
For the first time, DORA examined availability as a key
operations than their highest-performing counterparts.
measurement of software performance, in terms of both
knowing exactly what software will be available, when, and
making sure it’s accessible by end users. DORA found that
➎ Key technical practices drive high performance.
elite DevOps performers are 3.55 times more likely to have
These include monitoring and observability, continuous
strong availability practices.
testing, database change management, and integrating
security earlier in the software development process.
➋ So does your approach to cloud infrastructure.

The cloud improves software delivery performance, and Read the full report here.
teams that leverage all of cloud computing’s essential
characteristics (such as self-service functions, resource
pooling, and automatic scaling) are 23 times more likely to be
high performers.
*DORA was acquired by Google as of December 19, 2018; for more information, see announcement.

Real-time insights from streaming data
Cloud computing also makes big wins archived approximately 5 to 7 million of
possible at each stage of the data maturity its old photos in hundreds of file cabinets
In the streaming enterprise,
curve, allowing companies to use more of three stories below street level near its
their data, more effectively. Companies that Times Square offices in a location called
cloud’s real promise lies in
store data in the cloud are able to access the “morgue.” To preserve this priceless its ability to unite batch and
a greater share of their data, faster. For history, The Times is digitizing its archive, streaming data inputs to
example, Telegraph Media could process using Google Cloud Storage to store high- run real-time analyses using
up to 4 TB of analytical data in less than resolution scans of all of the images in
machine learning and AI.
a minute, which enables the tools they the morgue. Once this data is stored in the
use to provide more relevant content (see cloud, The Times uses services like Google
sidebar for more: “The future of news”).21 Cloud’s Vision API to identify objects,
Additionally, advanced analytics tools places, and images so reporters and editors
empower data scientists to quickly query can quickly (and cost-effectively) find what
structured and unstructured data from they need to tell a richer story.
diverse sources. Cloud platforms further
extend the value of a company’s data by In the streaming enterprise, the cloud’s
connecting it to other cloud services, such real promise lies in its ability to unite batch
as machine-learning-trained API models. and streaming data inputs to run real-time
analyses using machine learning and AI.
Cloud technology isn’t just for digital IoT devices, customer interactions, social
natives. It’s helping longtime industry media, and environmental data – to name
leaders like The New York Times turn just a few – all provide signals that, when
offline artifacts into valuable digital data. properly analyzed, can create a single
For more than 100 years, The Times has view of a customer, product, or interaction.

With the right tools and systems in place, “For ride hailing alone, we have more than
streaming enterprises can apply these 1 million drivers on our platform, and
signals in real time to drive smarter on a typical day we’ll have hundreds of
decision making, richer personalization, thousands of drivers online at the same
predictive maintenance, or proactive time, servicing our customers,” says Willem
security protections. Pienaar, data science platform lead at
GO-JEK. “We also have more than 300,000
GO-JEK, a technology startup based in merchants on our food delivery platform,”
Jakarta, Indonesia, that specializes in ride he adds. “We call them merchants because
hailing and logistics, offers an example of they are not just restaurants; they are
how the streaming enterprise unfolds – and mothers and fathers selling food from
its impact on the overall business. Founded their garages. Our platform supports
as a call center for ojek (motorcycle taxis) socioeconomic mobility in Indonesia by
bookings, GO-JEK used Google Cloud allowing people to rise up from poverty.”
Platform and Google Maps Platform to
establish itself as a leading on-demand GO-JEK has relied heavily on the skills and
multiservice platform and one of the few expertise of its technology team and on
unicorn businesses in Southeast Asia. selecting the right technologies to grow and
expand into new markets. Google Cloud
In 2015, the company launched a mobile Platform solutions play an integral role in
application that bundles ride hailing, food enabling GO-JEK to capture and analyze the
delivery, and grocery shopping. GO-JEK has data associated with 2.5 million customers
now bundled more than 18 products, one and 1 million drivers across 167 districts
loyalty program, and one electronic money and cities.
service into its application.

Streaming insights to feed the world
Growing up on a farm in Israel, Ofir Schlam would rise at With 19,000 farms around the world using this data-intensive
dawn to check the fields for signs of pink bollworm, an premium service, Taranis needs to ingest and process an
invasive species known for devastating cotton crops. His enormous volume of image data with the capability for near-
early days spent at the farm inspired him to found Taranis, real-time analysis. “Farmers can’t wait a week to get results,
an agro-tech company that takes care of such chores while when [the blight] has already expanded and caused a lot of
farmers are still fast asleep. damage to their crops,” Schlam says.
Schlam may joke that he was trying to avoid waking at 5 a.m., Taranis uses Google Cloud Platform to manage its image
but Taranis’ true genesis stems from a far weightier concern: database, as well as AI tools available on GCP for data
solving world hunger. With a growing world population, analysis. It’s a business that wouldn’t exist without the scale,
demand for food has never been higher, yet contamination speed, and real-time processing capabilities of the cloud
from chemical leakages, pests, and disease increasingly – a true streaming enterprise. Taranis has already helped
threaten vital food sources. Each year, an estimated one- reduce crop loss by more than 15 percent, and this is just
third of the food produced for human consumption is lost or the beginning. As cloud computing continues to evolve,
wasted, according to the Food & Agriculture Organization of Schlam has an eye out for new opportunities, including
the United Nations.22 In developing countries, much of that edge processing to enable not only real-time analysis but
waste and loss occurs at the early stages of the food value also automated responses (such as a drone that can spray
chain. In 2015, Schlam and his co-founders started Taranis a plant as soon as it detects a pest infestation). As Schlam
to help farmers assess and mitigate crop loss from disease, puts it, “There’s a lot more to innovate.” See how Taranis
insects, weeds, or nutritional issues. Today, Taranis collects saves the world’s food supply with cutting-edge AI.
images of 20 million acres of farmland with a sub-millimeter
camera mounted on a drone that can capture details as tiny
as a bug on a blade of grass in real time.

Top 3 tactics to build customer trust,
Security & trust according to a Google study24
Depending on whom you ask (and on their IT infrastructure in the cloud, Adkins
familiarity with cloud technology), security says, companies can scalably and
is either a top driver of cloud adoption, a top continuously manage every single asset, ➊ Build reliable computing services (85%*)
barrier to cloud adoption, or both.23 Bottom user, and configuration.
line, it’s top of mind – and cloud computing
Take Ravelin, a fraud detection provider
85%
holds huge promise for those looking to
manage security and trust more effectively. that flags fraudsters within milliseconds to
online merchants using real-time behavioral
Heather Adkins, Director of Information analytics. Because fraudsters are constantly
Security and Privacy at Google, says evolving their methods, Ravelin doesn’t
➋ Invest in security (84%*)
companies must first think of security in a just demand exceptional speed, security,
more fluid way. “The boundaries in security and scalability from its cloud provider, but 84%
have vastly changed,” Adkins explains. openness and adaptability, too, so it can take
“We used to imagine the boundary around advantage of open-source advances and stay
a simple physical computer or room of one step ahead.
computers. Now, as people carry their
computers in their hands all around the “We believe in open source and open ➌ Enable transparent communication (82%*)
world, we are transitioning to thinking standards and open protocols,” says Leonard
about how the boundaries have to be Austin, the company’s Chief Technology
82%
around the data.” Officer. “At Ravelin, we rely on a number
of open-source technologies to protect
Cloud computing makes security our clients from fraud. With Google Cloud
management possible in a porous and Platform we can do that in a fast, globally
varied ecosystem of platforms, users, scalable, and more secure way.”
and devices. By centralizing data and
*% of professionals who rated this option as “very

important” or “somewhat important”

Culture
Realizing the promise of cloud computing But cultural change requires leadership, and Meckfessel employs a specific methodology for
requires a change not only in how teams work scaling Google’s model. When asked about her guidelines, she recommended the following:
together, but also in how they think about their
work. For example, thinking about product
➊ Use data to guide decisions so all the humans are looking at all the same data.
development in a cyclical (versus linear) way
can help engineers see the bigger picture and
worry less about small mistakes. ➋ Be transparent; don’t hold information back.
“When people feel like they have each other’s

back – or that the tools, the continuous ➌ Collaborate so developers and operators have shared goals.
integration/continuous delivery, has their back
– they’re more likely to take risks,” Meckfessel
says. “They’re more likely to create; they’re more ➍ Bake in blameless post-mortems and accept that writing software has risks
likely to move faster.” and defects.
In this context, automating the mundane tasks isn’t just about making IT more productive. “It’s
data, transparency, collaboration, and blameless post-mortems because that’s how you build
trust, and that’s how you have fun together as a team. Software engineering is a team sport,”
Meckfessel says.

PART 4
How should
leaders
prepare?
In Part 1, we explored the enterprise of the future. In Part 2, we looked at how the
characteristics of cloud computing will shape the enterprise. In Part 3, we examined
cloud computing’s impact on the business as a whole. In this section, we’ll offer
actionable advice for leaders looking to seize the opportunity to transform.
STEP 1
Find the cloud model that works for you.

By 2022, the top four cloud “megaplatforms” Cloud’s CTO. “The modernization happens
will host 80 percent of infrastructure as a naturally as they increasingly integrate cloud
“All the machines we’re offering, all
service (IaaS) and platform as a service services into their IT toolkit. It’s become not an
(PaaS) workloads, according to IDC.25 As cloud if or a why; it’s become a how.” of this technology, all of this code –
adoption increases, hybrid and multi-cloud it’s really about trying to maximize
models will become the norm, with every As CTO, Stevens works directly with the amount of human creativity we
enterprise assembling a mixture of cloud companies that are figuring out their “how.” can bring to bear.”
services and legacy applications to meet their Google’s model is based on a partnership:
needs. This approach brings the cloud to the Google technical teams sit with the customer’s
enterprise, rather than bringing the enterprise IT team to work through a new process, – Melody Meckfessel,
to the cloud. We can expect a world in which step by step. Recently, Stevens saw one of VP Engineering, Google Cloud
cloud computing is the standard – and where these partnership teams migrate three major
open, interoperable tools play a crucial role in enterprise resource planning workloads to
connecting a company’s data and applications the cloud in the space of two weeks. (The
across environments. company’s leadership, he says, was “stunned...
people think it’s harder than it is.”) “The idea
“Organizations have seen, from their is, when [the Google engineers] leave, things
experience in the public cloud, that they don’t stop,” Stevens says. “Next workload,
can develop faster, iterate faster, and bring next workload, and then that just creates a
solutions out faster,” explains Stevens, Google snowball effect within the organization.”

STEP 2 Open-source software (OSS)
establishes a seat at the table26
Build on an open platform.
The benefits of open-source software are clear: interoperability across on-
premises and multi-cloud environments, protection from vendor lock-in, and
continuous innovation from a community of third-party developers. As cloud
of companies plan
computing more deeply penetrates every company’s IT stack, choosing an 94% to use OSS by 2029.
open, interoperable platform becomes more important than ever.
“Open, emerging technologies of today, like Kubernetes, in 10 years will

become dial tone,” Stevens predicts. “It’s not sufficient that [public] clouds
have their proprietary way of doing things. We need them to all work well
together – that needs to be our north star.”
of companies plan
Open source has played a pivotal role in enabling standardization and to use OSS for the
interoperability across cloud providers and other technologies. In addition 41% majority of their
to incorporating OSS into the enterprise stack, Stevens advises companies software platform
to look for cloud partners that prioritize openness – by open-sourcing their by 2029.
own innovations, contributing to the community, or designing platforms with
interoperability in mind.
“If we built the best cloud for containers, because we invented the container
model and orchestration with Kubernetes, it’ll just mean we’ve got a great
cloud. We won’t have changed the industry,” he says. “The open-sourcing of
Kubernetes, Istio, and Tensorflow is the path to helping the whole industry
move along together.”

STEP 3 Automation to speed up security
Rethink IT security.
of decision makers
Currently, many companies approach IT security with a pre-streaming expect a majority of
mentality: protect the perimeter, patch vulnerabilities as quickly as possible, 70% security operations
and go back to hoping the perimeter won’t be breached again. Adkins advises to be automated in
companies to turn this model on its head: assume constant vulnerability, and 10 years.27
you’ll never be surprised.
“Just as we’ve started thinking about streaming medicine (always-accessible

healthcare personalized to our bodies), we should think about IT streaming
security on a constant basis,” Adkins advises. “Development will need
Security is shifting to the left
to move into a continuous feedback model so that code can be pushed
(refreshed, fixed, and updated) at any time. The benefit is that if there is
a security issue, it can be dealt with immediately as opposed to waiting of decision makers
quarterly, yearly, or any other time down the road. This will reduce and expect to see
eventually get rid of the periods of vulnerability when an adversary can more security
exploit the company.” 72% implementation
pre-development
Adkins also emphasizes the importance of continuous learning and a
by 2029.28
no-blame culture in which it’s OK to make mistakes as long as they’re
exposed, studied, and fixed for the long term. She thinks of security as “a
workplace revolution” in which an innovative IT department dedicated to
running advanced data analytics at scale can be a critical force in protecting
a company’s data and customers.

STEP 4
Lead the change.

As with any new approach, leadership and change management are
critical factors in the new model’s success. A recent survey from the
Harvard Business Review found the vast majority of decision makers (83
percent) believe that a change-embracing culture is extremely important
to their organization’s ability to successfully adopt new digital technology
capabilities quickly.29
For most companies, making the transition to the cloud will require not
only rethinking talent strategy and organizational structure, but also a
reconsideration of a company’s overall culture.
At Google, Meckfessel describes a shift “to really connect everything back

to people... helping people realize their ideas, helping people create.” As
access to computing resources grows and cloud technology penetrates new
markets, she adds, diversity plays an increasingly important role in product
development (see sidebar, “The case for diversity”).

And while Meckfessel’s focus is on developers and operators, even those
definitions are expanding. By 2024 a new class of professional developers
“There’s something magical in understanding
producing code without custom scripting will expand the developer population
what makes people productive, helping them
by 30 percent, according to IDC30 – further expanding what it means to build
on cloud platforms in the streaming enterprise. with that, collaborating with them, and then
they are empowered to influence themselves.”
“There’s something magical in understanding what makes people productive,
helping them with that, collaborating with them, and then they are empowered
– Melody Meckfessel, VP Engineering,
to influence themselves,” Meckfessel says. “The culture for the developers
Google Cloud
and operators is a significant part of what makes them happy, productive, and
focused on those business outcomes.”
Air cover from senior leadership is crucial, Stevens adds, particularly when the
changes may be viewed by some as disruptive. He also advocates securing
easy wins wherever possible: “Pick some initial workloads, create a few tiger
teams, bring in an experienced partner, and build toward a positive outcome
measured in weeks rather than quarters.”

The case for diversity
Compute is becoming more affordable. developing these solutions for. The path
Ease of access to compute is increasing toward that is not just what we offer the
across different communities and different humans within Google Cloud, but that we
geographies all over the world. We’re at a have an opportunity to enable inclusive,
very interesting point in cloud computing respectful behaviors for diverse software
where we’re on a journey with customers that developers and operators within the software
are migrating to the cloud. development lifecycle. That will continue to
increase the level of conversation, the sharing
What I think is happening and what we’ll see of ideas – whether it happens in code review
happen in the next five to 10 years is as cloud or if it happens earlier on in the software
technology offers more solutions to our development process – that will lead to better
customers, we’re going to see different types products for users out in the market. Melody Meckfessel
of developers coming into cloud computing.
There is an openness and an access that is Developers have different requirements VP of Engineering,
going to open up a diversity of educational and constraints that they’re working in, and Google Cloud
backgrounds for developers. The number one at Google Cloud we want to offer them the
job [in 2030] is going to be for developers. tools, the underlying platform components,
to be able to do what they need to do within
The diversity of who’s creating the product whatever segment, or identity, or community,
for end users has not represented the user or ecosystem that they’re living in, and we
community. There have been some real know that those communities change over
misses, and I think we need to do a better time. And the identities of developers and
job of representing the users that we’re operators change over time.

Conclusion
Increasingly, technology, data, and human interactions will follow a new model of
real-time inputs, enabling change without interruption and positive iteration. Cloud
computing powers an important shift for information technology, and it will usher
in new and better ways to serve customers, make discoveries, and build great
products and services.
As part of this transformation, technology models will become more flexible,

interoperable, and open; office cultures will shift toward transparency, collaboration,
and constant learning. That means innovation and team building – the best parts
of work – can happen more frequently and efficiently. Customer focus can sharpen
and become more responsive. Corporate knowledge gained over decades of
competing and innovating can be brought to bear more easily and meaningfully.
To compete effectively in this changing landscape, business and technology

leaders must find ways to embrace the potential of cloud computing – whether
in the tools and systems they choose, the institutional cultures they create, or the
business strategies they prioritize. Today’s enterprises can set the stage for future
success by infusing their unique assets – such as customer relationships, excellent
service capabilities, strong partnerships, market knowhow, and technology
investment – with the power of innovative IT.
We’re excited about the future, and we’re eager to partner with companies to share
our vision of openness, excellence, and innovation.
Get in touch

Appendix
1. RightScale, 2018 State of the Cloud Report, January 2018 8. Google Internal Study analyzing predictions and trends in cloud computing among
IT and business decision makers at global mid-market (500 to 999 employees)
2. IDC FutureScape, Worldwide IT Industry 2019 Predictions, November 2018 and large companies (1,000+ employees), January 2019
3. IDC Market Forecast, Worldwide and Regional Public IT Cloud Services Forecast, 9. Google Internal Compute Study, January 2019
2018–2021, March 2018
10. Google Internal Compute Study, January 2019
4. Smarter With Gartner, “What Edge Computing Means for Infrastructure and
Operations Leader”, October 3, 2018 11. Google Internal Compute Study, January 2019
5. IDC FutureScape, Worldwide IT Industry 2019 Predictions, November 2018 12. IDC White Paper, sponsored by Seagate, The Digitization of the World From Edge
to Core, November 2018
6. IDC FutureScape, Worldwide IT Industry 2019 Predictions, November 2018
13. Cisco White Paper, Global Cloud Index: Forecast and Methodology, 2016–2021,
7. IDC Market Forecast, Worldwide and Regional Public IT Cloud Services Forecast, November 2018
2018–2021, March 2018
14. MIT Technology Review Custom and Google Cloud, Machine Learning: The New
Proving Ground for Competitive Advantage, 2017

15. The New Stack, Survey: Open Source Programs Are a Best Practice Among Large 23. Mckinsey & Company, Making a secure transition to the public cloud,
Companies, August 2018 January 2018
16. IDC FutureScape, Worldwide IT Industry 2019 Predictions, November 2018 24. Google Internal Compute Study, January 2019
17. Google Internal Study analyzing adoption pathways to public cloud through 25. IDC FutureScape, Worldwide IT Industry 2019 Predictions, November 2018
containerization for the Enterprise segment (January 2019)
26. Google Internal Compute Study, January 2019
18. Harvard Business Review Analytic Services, Competitive Advantage through
DevOps, January 2019 27. Google Internal Compute Study, January 2019
19. Google Internal Compute Study, January 2019 28. Google Internal Compute Study, January 2019
20. Harvard Business Review Analytic Services, Competitive Advantage through 29. Harvard Business Review Analytic Services, Leading in a world of continuous
DevOps, January 2019 change, 2018
21. https://cloud.google.com/customers/telegraph-media-group/ 30. IDC FutureScape, Worldwide IT Industry 2019 Predictions, November 2018
22. Food & Agriculture Organization of the United Nations, Key facts on food loss and
waste you should know!

Future Of: Cloud Computing

Uploaded by

Copyright:

Available Formats

Future Of: Cloud Computing

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Future Of: Cloud Computing

Uploaded by

Copyright:

Available Formats

Future of

A stake in the future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

Part 1: The enterprise of the future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

How cloud computing will change the way we work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

Part 2: Fundamentals of the cloud enterprise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Future of cloud computing

Leadership & culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

Part 3: What are the impacts for business? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

Speed & agility at scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

Real-time insights from streaming data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

Security & trust . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

Future of cloud computing

Step 1: Find the cloud model that works for you . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

Step 2: Build on an open platform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Step 3: Rethink IT security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

Step 4: Lead the change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

Future of cloud computing

Future of cloud computing | 5

Google’s mission is to organize the world’s information and make it universally

Future of cloud computing | 6

– Urs Hölzle, SVP Technical Infrastructure, Google Cloud

Future of cloud computing | 7

Future of cloud computing | 9

If the value of a technology is the net experience it delivers to the enterprise

In the following pages, we’ll explore:

➊ What the enterprise of the future will look like

➋ What fundamentals shape the cloud enterprise

➌ How cloud computing will impact businesses

➍ How business and IT leaders and decision makers can prepare

Future of cloud computing | 10

Future of cloud computing | 11

the way we work 77%

In technology security, elements like BeyondCorp, Google’s zero-trust enterprise

*IT and business decision makers at global

Future of cloud computing | 12

➊ Collaborative technologies like document sharing end the stop/

Future of cloud computing | 13

➍ Even slow-changing aspects of business, like supply chain and

Future of cloud computing | 14

Future of cloud computing | 15

In Cerf’s view, better products stem from a clearer understanding of how

Future of cloud computing | 16

When it comes to investing in enterprise software, most companies are not

Future of cloud computing | 17

Future of cloud computing | 18

An increasingly important dimension is the advent of cloud-type technologies

In this section, we’ve taken a technology-driven approach to analyzing how

*Global mid-market and large companies were surveyed.

Future of cloud computing | 20

To meet these challenges, companies are turning to the cloud. By moving

Future of cloud computing | 21

Learn more about our data center approach at

Source: DeepMind blog

Future of cloud computing | 23

Future of cloud computing | 24

Future of cloud computing | 25

Future of cloud computing | 26

Future of cloud computing | 27

“[Migration to the cloud] doesn’t necessarily start by changing the whole