DevOps in the Internet of Things

Six Reasons It Matters and How to Get There

WHEN IT MATTERS, IT RUNS ON WIND RIVER

 DEVOPS IN THE INTERNET OF THINGS

EXCUTIVE SUMMARY

It’s increasingly apparent that the development of software for the Internet of Things

(IoT) and the management of those systems once they are in operation cannot be sepa-

rated—making IoT software an area ripe for “DevOps.” More than a buzzword, DevOps

has the potential to help accelerate system development, ensure system quality, and

optimize system reliability in the field.

DevOps is already being used in the IoT enterprise systems where the business logic

resides. This paper sets forth six sound reasons why DevOps should, and likely will,

become standard practice in the development and deployment of software for gateways

and edge devices as well, and outlines a technology infrastructure that can help organiza-

tions implement IoT DevOps more quickly and easily.

TABLE OF CONTENTS
Executive Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
The Evolution of DevOps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Six Reasons DevOps Matters in IoT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Overcoming the Obstacles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
How Techology Can Enable DevOps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2 | White Paper
 DEVOPS IN THE INTERNET OF THINGS

THE EVOLUTION OF DEVOPS of an IoT solution requires a constant feedback loop, regular
The integration of development and operations, dubbed monitoring, speedy issue resolution, and frequent upgrading.
“DevOps,” is a hot topic in IT circles—so hot, in fact, that a stan- Internet connectivity creates the opportunity for constant infusion
dard definition, let alone formalized DevOps structures and best of innovation into the system, without waiting for the next “big
practices, is yet to emerge. One source characterizes DevOps as bang” release. It’s a process of continuous learning that necessar-
a “culture, movement, or practice” emphasizing collaboration ily requires developers and operators to collaborate closely every
between developers and IT operations teams with the goal of day.
creating an “environment where building, testing, and releasing Agile development is the forerunner to DevOps. In agile devel-
software can happen rapidly, frequently, and more reliably.” opment, designers, testers, developers, and integrators merge
Note that this definition refers to “culture” rather than organiza- into cross-functional teams that have end-to-end responsibility for
tion. While it’s true that DevOps may ultimately require an organiza- specific functions or subsystems (see Figure 1). That responsibility
tional change, it first requires a cultural change to break down silos includes delivery, which ideally can be automatic once the soft-
that separate those who build software and systems from those ware passes internal testing and quality assurance. Automation of
who implement and operate them. (The evolution of DevOps may delivery makes possible the concept of continuous deployment,
be likened to that of agile development, which began as a move- which increasingly goes hand in hand with agile methodologies.
ment with the “Agile Manifesto” and is still thought of more as a If your organization practices agile development and continuous
set of principles than a process.) deployment, you are on the evolutionary path to DevOps. You may
For developers of embedded systems, the concept of DevOps even be practicing DevOps in an ad hoc fashion without realizing
may at first seem foreign. Historically, the development team built it. The next step is to formalize a DevOps organizational process
the software (or a device) and handed it off to another team for and structure, supported by technology that integrates the build-
release and support. But in IoT that model is a recipe for some- ing, testing, deployment, and management of IoT applications on
thing far short of success, if not failure. The reliable performance a single platform with a high level of automation. But before we
discuss how you do it, let’s be clear about why you should.

Functional Organization Cross-Functional Organization

Team Team Team Team End-to-End

Responsibility
Team Design Design

Team Develop Develop

Team Test' Test'

Team Integrate Integrate

Team Deliver Deliver

Team Operate Operate

Figure 1. Functional vs. cross-functional organization

3 | White Paper
 DEVOPS IN THE INTERNET OF THINGS

SIX REASONS DEVOPS MATTERS IN IOT accelerates development cycles without compromising quality
DevOps has already gained a foothold at the enterprise level. through more effective collaboration is simply a better, faster,
Witness the social media networks or video or music streaming smarter way to work—with the potential to drive down operat-
services that are constantly pushing new content to users. Or ing costs.
consider the mobile device makers and application providers that The case for evolving DevOps from enterprise systems to IoT edge
make new releases of operating systems and other apps available devices is fairly compelling. IoT application and device developers
for download on a regular basis. Now, the next frontier for DevOps are under enormous pressure to deliver quality solutions and meet
is IoT, specifically the edge devices (the “things” in the Internet tight time-to-market demands. And because IoT systems may be
of Things) that perform the IoT system operations and feed data expected to perform for many years, development and operations
back to the enterprise. Here’s why: teams must work together to plan for their entire lifecycle, from
• The evolution effect: DevOps is simply bound to happen, in design through end-of-life. Companies that can meet these chal-
some shape of form, as the next step in the evolution from agile lenges stand to gain a significant competitive advantage—and a
development to continuous deployment. Organizations that transition to a formalized DevOps organizational structure would
embrace DevOps and formalize it through integrated, cross- seem to be the answer for adapting to this new environment.
functional teams are likely to have a decided advantage over So why isn’t it happening more quickly? Let’s look at some of the
those that do not. obstacles and challenges to DevOps implementation.
• The spreading effect: If software on enterprise servers is being
updated regularly or continuously, the systems that are con- OVERCOMING THE OBSTACLES
nected to those servers and dependent on that software will Change is rarely easy, and instituting DevOps is no exception.
likely require frequent updating as well. At some point, teams Organizations are likely to encounter several obstacles in the
that develop the enterprise software will expect faster release transition.
cycles in the other parts of connected systems.
• Cultural and organizational change is difficult: A transition to
• The infrastructure in place: With systems now connected via the
DevOps entails overcoming years of ingrained cultural percep-
Internet and the cloud, it’s possible to automatically deploy and
tions and behavior. Changes in reporting relationships, respon-
regularly upgrade software in multiple field devices remotely.
sibilities, and accountabilities are bound to be a bit rocky. An
• Software-defined “anything”: Increasingly, it is the software
organization must recognize the need to change and then
deployed on a device (regardless of the hardware) that differ-
build the processes and systems necessary to accomplish the
entiates it and defines its functionality. That means that when
DevOps vision.
functionality needs to be updated, it is more often going to
• DevOps is inherently difficult at the device level: Unlike the
entail a software update rather than an electrical or mechanical
enterprise software environment, where server hardware is
modification.
fairly standardized, IoT systems can be very large, unique, and
• New business models and revenue streams: The big promise of
complex, with a wide variety of hardware platforms. Where
IoT is that it makes possible new business models and sources
enterprise systems have built-in redundancy, there is typically
of revenue that couldn’t otherwise exist. The ability to con-
very little redundancy for software embedded in field devices.
stantly deliver new software updates makes it possible to sell
Reducing the risk of costly failure requires exhaustive produc-
services that generate continuous revenues, rather than simply
tion-level testing and quality assurance, which lengthens devel-
the one-time sale of the underlying product.
opment cycles.
• Greater productivity and cost-efficiency: A process that
• Reliability is paramount: When software is continuously

4 | White Paper
 DEVOPS IN THE INTERNET OF THINGS

deployed, it has to work as promised. IoT solutions have

very high demands for reliability, quality, security, and safety.
Correcting problems in deployed software can be extremely
cumbersome and inefficient, and the business risk is high if the
manufacturer has a contractual obligation to ensure perfor- Development
mance as expected. Quality assurance is an essential ingredient (Software Quality
of any DevOps model. Engineering) Assurance (QA)
There are few tools that actually support the DevOps paradigm, DevOps
which calls for agile code sprints, automated testing, fast and
automated feedback loops, and collaborative teams with a high
degree of autonomy and communication. What’s needed is a clear
path between development platforms and field systems, so that
DevOps teams can monitor system health, detect potential issues,
and act on them before they become problems. Technology
Operations
HOW TECHNOLOGY CAN ENABLE DEVOPS
DevOps in IoT is not inevitable. It requires commitment, col-
laboration, communication, and a willingness to change. It can,
however, be made easier with technology that integrates system
Figure 2. DevOps is the intersection of development, QA, and operations
development, testing and debugging, deployment, monitoring,
and management on a single platform. Unlike conventional devel- • Wind River Helix App Cloud serves the development side of
opment tools designed to support functional organizations work- DevOps. It equips application developers with ready-to-use
ing in horizontal layers, an integrated approach would enable true tools and software development kits (SDKs) for any hardware
cross-functional collaboration in a vertical model (refer to Figure variant. App Cloud makes it possible to easily build applications
1). It would allow system developers and those responsible for independent of device operating system and hardware com-
operationalizing the software to work as a coordinated team in plexity. By providing developers with the appropriate tools and
a centralized, cloud-based environment, thereby accelerating the target systems, App Cloud helps mitigate integration issues
delivery of applications with full quality assurance and enabling between application and platform software and cut down on
effective troubleshooting of systems in the field. team handovers. As a cloud platform, it also allows anytime,
anywhere access to tools and enables large, geographically dis-
Wind River® Helix™ Cloud is an example of such a solution. It’s
designed to help teams build and deploy IoT systems more persed development teams to collaborate across borders and

quickly and manage them more effectively, leveraging the cloud time zones.

for anytime, anywhere access and enhanced collaboration. • Wind River Helix Lab Cloud addresses the testing and quality
assurance aspect of DevOps. It allows instant access to virtual
It’s important to think of DevOps not simply as the merger of hardware of whole systems at representative scale, enabling
development and operations, but as the intersection of develop- teams to use full-system simulation for testing and QA of com-
ment, quality assurance (QA), and operations (see Figure 2)—QA plex and large-scale IoT systems. A simple login provides any
being the essential step that ensures a system will work properly engineer with access to the cloud-based virtual lab. Using on-
before going into operation. Helix Cloud maps to this three-stage demand simulation software as a complement to hardware,
DevOps model through the integration of three core components: teams can automate testing in entirely new ways and create

5 | White Paper
 DEVOPS IN THE INTERNET OF THINGS

any number of virtual target systems for parallel testing. This CONCLUSION
significantly shortens the cycle between application develop- Connected IoT edge devices have a lifecycle beyond “deploy,
ment and system testing. With Lab Cloud, teams can stage and break and fix, and retire.” Connectivity creates the opportunity to
test systems at representative scale in a pre-production envi- continuously infuse incremental innovation across the system life-
ronment so they can move into production with confidence in cycle. DevOps puts organizations in the best position to capitalize
system continuity. on that opportunity.
• Wind River Helix Device Cloud is the platform for managing
As the lines between software creation and operation begin to dis-
and operating devices from the cloud—the “bridge” between
appear, so too must the organizational lines that separate system
development and operations. It enables operators to safely and
developers from system operators. The DevOps concept has been
securely update, monitor, service, and manage devices in the
proven at the enterprise level and is evolving toward gateways
field. Device Cloud automatically collects and integrates data
and edge devices as a means to meet the unique challenges of
from hundreds or thousands of disparate devices, machines,
developing and managing IoT systems. While there are obstacles
and systems, enabling operators to track device status and con-
to overcome, both cultural and practical, a technology platform
tent, share data among engineers, and proactively determine
that integrates system development, testing and QA, deploy-
when updates are needed.
ment, and management can provide the necessary infrastructure
Collectively, the Helix Cloud suite enables organizations to transi- for implementing DevOps, empowering cross-functional teams to
tion into a DevOps team structure. Specifically, it: accelerate system development, ensure system quality, and opti-
• Facilitates a new paradigm of “always connected” collabora- mize system reliability in the field.
tion that isn’t restricted by geography
• Helps break down the silos that separate those who develop
software and systems from those who deploy and operate
them, enabling teams to work cross-functionally in a collabora-
tive spirit and making the release of software fast, reliable, and
automated
• Allows testing of software at scale before deployment as well as
proactive management of field devices, which together enable
continuous quality assurance

Wind River is a global leader in delivering software for the Internet of Things. The company’s technology is found in more than 2 billion devices, backed by world-class professional services and
customer support. Wind River delivers the software and expertise that enable the innovation and deployment of safe, secure, and reliable intelligent systems.
©2016 Wind River Systems, Inc. The Wind River logo is a trademark of Wind River Systems,Inc., and Wind River and VxWorks are registered trademarks of Wind River Systems, Inc. Rev. 06/2016