4UBUF PG OEVTUSJBM OUFSOFU QQMJDBUJPO …predix-io-assets.s3.amazonaws.com/2016 State of Industrial Internet... · almost two-thirds of indus-trial developers apportion more than

2016: State of Industrial Internet Application Development

Sponsored byGE Digital

Evans Data Corporation340 Soquel AvenueSanta Cruz, CA 95062

800-831-3080www.evansdata.com

© 2016 Evans Data Corporation | January 2016 2

State of Industrial Internet Application Development

Study Highlights

This study, carried out in collaboration with GE Digital,

surveyed the existing industrial developer landscape, to better

understand who industrial developers are, how they allocate their

time and resources when developing applications, the challenges

faced in the development process, and the technological

opportunities available to them. The study, a survey of over 1,200

industrial developers, concludes that there is a need within the

industrial developer community for focused tools and that these

developers would receive significant benefit from using PaaS and

infrastructures such as Predix. Relevant findings include the

following:

Non-development tasks consume industrial developers’ time:

Unfortunately for optimized productivity, developers spend much of

their time on tasks that are related to development but that don’t

directly contribute to production of a finished application. On

average, industrial developers spend only 26% of their time writing

application source code, with the rest being spent on environment

tasks such as optimizing for security, availability, scalability and stack

integration.

Platform as a Service addresses productivity issues: PaaS

solutions leveraging the Cloud minimize developers’ need to create

and administer a development environment, thus industrial

developers are embracing developer PaaS solutions. Over 70% are

currently using a PaaS or evaluating one .

Major challenges still exist: Although PaaS offerings provide

many solutions, industrial developers still see looming challenges for

building Internet of Things applications. These include security,

communications, access to specialized target devices and data

streams, and lack of standards, among others.

Security is a major issue: Industrial developers, like developers

across the board, perceive security as the top threat in the

development and implementation of the Industrial Internet of Things.

In our study this was the number one issue cited by developers, with

the plurality (37%) mentioning security.

Analytics is a necessary adjunct to development: While very

few industrial developers spend the majority of their time working

with analytics, it plays a part in virtually all of their development

efforts. The plurality spends between 25 and 49% of their

development time working with data analytics and only 12% of the

developers spend less than 5% of their time on this.



Productivity and Challenges

NoSQL is gaining on data historians: Although data historians and SQL databases are currently better used

for time series data management in the industrial sector, more industrial developers project that they will be using

NoSQL in the next 18 months than anything else.

Industrial developers target multiple disciplines: While the majority of industrial developers currently or

planning to target specialized industrial machine controls, they also target many other types of devices, including

wearables, automotive, consumer boards, robotics and home devices.

GE’s Predix PaaS is a robust software platform designed to address the specialized needs of industrial

developers, while also offering flexibility that can be used across device and implementation types. Almost three-

quarters of the industrial developers surveyed have tried or have plans to evaluate Predix. This study explores the

ecosystem for industrial development that Predix will flourish in.

Although the modern-day industrial world is rife with potential, there are always issues to contend with

and as new technologies arise, their adoption demands new platforms and tools to help developers address them.

Industrial developers today point to the complexities of the Industrial Internet environment and the fact that they

are still learning how to navigate it and effectively use its resources. Ideally, developers should spend a majority of

their time on task: writing application code. However, that’s not the case for industrial developers, with almost

75% of the time spent on tasks other than innovation.

There is change in sight, though. The advance of cloud-based and Platform as a Service (PaaS) offerings

promise to free up developers to focus more of their efforts on actual application code development, while

enabling more rapid prototyping and more frequent deployments into industrial productions.

Further, adoption of modern development disciplines like DevOps, automated build/test and continuous

integration allows developers to further leverage today’s cloud platforms to enhance and enable more rapid and

coordinated deployment.

While data historians and SQL databases have dominated data storage and time series management in

the past, there is evidence in this study that a future emphasis on NoSQL may come into play.

Platforms for industrial development must assure developers of built-in and robust security, especially for

Industrial Internet of Things development where security is a huge concern. But they also must supply APIs that

are relevant to the industry - and support the most commonly targeted devices and the most popular

communications standards.

The sum of the findings from this study suggests that the use of a cloud-based PaaS would greatly benefit

industrial developers. Not only would PaaS usage reduce the amount of time spent on non-value added activities

and facilitate collaboration between the various stakeholders in the development process, use of a robust but

focused PaaS (like Predix), would also provide a powerful analytical tool kit which unites, via the internet, the

various industrial connected devices and provides accurate, up-to-date information on the performance of these

devices.



Technologies and Resource Allocation Practices of Industrial Developers

One of the great challenges facing industrial developers is the suite of other job related responsibilities that

detract from time spent writing code. Very few industrial developers spend the majority of their time actively

designing or developing

applications. Instead their

time is often taken up by

technology administrative

duties such as optimizing

for data security, system

availability or scaling.

Hardware acquisition and

integrating technical stacks

also consumes time that

could be spent on actual

development.

Within the course

of development for any application, there are tasks that are must be done in addition to writing code. Developers

also need to design the structure of their applications, model the application flow and the structure of the data,

build, test and debug their code, and carry on regular maintenance for existing applications. However, the activities

we investigate here are those that are entirely peripheral to software development, such as provisioning the

hardware and middleware, setting up machine instances, and optimizing for security, availability and scaling. While

necessary, these types of tasks fall outside the purview of active development and are exactly the types of tasks that

can be performed by a competent Cloud service provider and are optimized in a PaaS offering.

Considering the average amount of time developers spend on

these cloud-centric tasks, the appeal of a PaaS solution becomes clear. The

table below shows the average time industrial developers spend on each

of a variety of peripheral tasks. While writing source code for applications

consumes 26% of their time, merely optimizing for security, availability and

scale takes up to 40% average.

Clearly this shows an industry ripe for the added value that cloud-

based PaaS systems offer. A PaaS solution takes care of all the administra-

tive issues of provisioning the hardware, integrating the technical stacks,

spinning up virtual machines and initiating new application instances. A

good PaaS solution provides optimized security, systems availability and

scaling.

Mean: What percent of your time at work do you spend on each of the following?

Valid Percent

Writing app source code 25.7

Procuring hardware & IT services 17.9

Integrating tech. stacks 14.1

Optimizing for data security 13.8

Optimizing for system availability 13.5

Optimizing for scale 12.4

Other 13.4 General Electric Industrial Application Development

Survey © 2016, Evans Data Corp.

Average Time Spent on Development Peripheral Tasks

Two of the major goals of this study are to 1) understand how developers allocate their resources currently,

and 2) to identify the challenges they face in the development process. This includes understanding the types of

development tasks that weigh on industrial developers and erode productivity, to see where their efforts that might

be more efficiently addressed through superior tools, and how new technologies can assist industrial developers and

make them more effective in their development efforts.



Time Spent on Data Analytics

Data acquisition and management is key in many industrial implementations. Measurements, data

recording and storage are often major factors in applications, but data is nothing without analysis. Thus industrial

developers find that much of their time, even when directly working on applications is involved with data

analytics.

In this study we asked the developers how much of their development time they spent working on data

analytics. In most industrial

processes the collection and

analysis of operational data

is a primary variable in the

optimization of the output.

Early detection of possible

failure of industrial assets or

process inefficiencies ena-

bles substantial business

performance optimization.

It is not surprising then that

almost two-thirds of indus-

trial developers apportion

more than 25% of their time

to data analytics.

Considering that the majority of industrial developers spend significant amounts of time on data analytics,

tooling that enables and aids in that analysis would have significant value in the industrial development

ecosphere.

Time Series Management

In factories, warehouses, plants and manufacturing facilities there are thousands of time series data that

coexist with other data types. Each has to be managed and often integrated. A time series collects observations

that are recorded or measured over time on a regular or irregular basis and for the most part sequentially.

Examples might be temperature, electrical charge, system wear, transactions, and so forth. Databases and data

historians are used for capturing and storing time series data.

Databases used in industrial applications vary in technologies used and database architecture, so a

further line of inquiry in our study looks at plans for using each of the three most used technologies: data

historians, NoSQL databases and SQL databases.

What percent of your development time is spent working with data analytics?

Valid Percent

Less than 5% 12.4

5-24% 24.8

25-49% 35.6

50-74% 18.8

75-99% 5.9

100% 2.6

General Electric Industrial Application Development Survey © 2016, Evans Data Corp.



Roughly 60% of industrial developers currently use SQL, with an additional 30% planning to adopt. SQL

databases are the most commonly targeted time-series data management tools. Interestingly, despite the tailoring

of data historian databases to the collection of industrial data, SQL databases are more often implemented.

NoSQL databases differ from relational databases, like SQL, in their tendency to stray from tabular

relations. In fact, to be deemed a NoSQL database, a database must not depend largely on tabular data storage

and retrieval architectures. NoSQL type databases have experienced a surge in popularity in the 21st Century. The

findings appear to corroborate this surge; industrial developers plan to target NoSQL databases more than any

other listed time-series data management program in the future.

Perc

ent

of

Pro

gra

mm

ers

General Electric Industrial Application Development Survey © 2016 Evans Data Corp

What are your plans for using the following for "time series" data management?

Currently target

Plan to tar-get within

the next 18 months

Plan to tar-get after 18

months No plans

Data Historians 43.4 23.6 5.6 27.3

NoSQL 33 34.2 8.4 24.3

SQL Databases 63 18.8 8.5 9.6

Others 18 16.4 9.7 55.9




DevOps, Continuous Integration and Build/Test Automation

In addition to examining how industrial developers allocate their development time, this study also sought

to identify how supporting disciplines and development ideologies are embraced by this specialized group of

developers. We asked developers how important development team coordination and production concepts like

DevOps, test/build automation and continuous integration/delivery were to their development activities.

Developers in general, have long been concerned with optimizing the development process from

beginning to end, so the adoption of these methodologies is no surprise. A large majority of developers,

approximately two-thirds of respondents, employ each of these techniques at least in a basic capacity, within their

organizations.

DevOps calls for collaboration between major stakeholders in the development process – most notably

development and operations. As a system of collaboration and cooperation between these two groups, DevOps

facilitates software production, testing, and maintenance and plays a role in shaping the progress within the

development lifecycle. The practice of DevOps is intended to maximize not only the quality of the application but

also ensure that the application can integrate well with targeted platforms, and that errors are not introduced by

variations in the development and production environments.

Test automation calls for the automation of quality assurance practices. Testing is done via a series of

scripts and is performed throughout the development life cycle so that comparisons can be made between

successive tests.


Perc

ent

of

Pro

gra

mm

ers



Plans for Using Platform as a Service (PaaS)

A lot of the daily non-development tasks that industrial developers perform and that erode productivity

can be eliminated with the use of a cloud-based PaaS. The PaaS takes care of all the hardware acquisition and

configuration, security, scalability and other administrative technical functions, plus it forms the underpinning to a

continuous delivery and DevOps strategy.

Considering the benefits that a PaaS has to offer, we asked industrial developers about their plans for the

cloud adoption and more

specifically, we asked their

plans for using the cloud in

a PaaS (Platforms as a

Service) capacity. We’ve

seen a rapid adoption of

PaaS within every sector of

software development and

industrial development is

no exception. Working

within a PaaS permits for

reduced complexity and

less time spent on non-

development tasks when

compared with traditional

non-cloud based platforms.

Moreover a PaaS facilitates

collaboration between

teams of developers.

Continuous integration refers to the process of coding, particularly in teams, and involves merging all

developer working copies of a code to a mainline/trunk copy several times a day and each check-in is validated

by an automatic build, thus allowing for the early detection of problems or bugs.

That industrial developers use these disciplines suggests that they are concerned about quality and

efficiency and that they can benefit from collaborative tools that facilitate exchange between the various

contributors. Data analytics tools that allow for robust quality assurance may be similarly valued.

How important are each of the following con-cepts to your development activities?

It factors into all my develop-

ment work

I use it frequently

I have used it in a basic

capacity

I am famil-iar but have not used it

I am unfamiliar

DevOps 18.8 22.9 20.4 16.8 21.1

Build/ test automation 20.3 27.1 21.5 17.4 13.6

Continuous integration /delivery 21.1 25.2 20.7 16.7 16.4




Internet of Things and Top Challenges for Industrial Developers

The Internet of Things (IoT) is already connecting computing devices, hardware, products, animals and

humans through the Internet, accumulating data that will then be used to improve a vast array of items and

experiences throughout the world. From environmental to industrial concerns, the Internet of Things provides a

goal for companies and vendors alike. Nowhere is the IoT flourishing more than in the industrial sector.

Other Evans Data surveys have shown that industrial applications are the top implementation for IoT

developers, and the adoption of IoT in industry is facilitated by the fact that rotating assets and control systems,

factories, warehouses, and plants already have in place sensors, automation of every kind, and in some cases

robotics. That IoT development would thrive in an industrial environment is no surprise.

But that doesn’t mean there aren’t issues involved with IoT in an industrial situation. To find out the top

challenges we included questions in our survey.

Almost three-quarters are either currently using or planning to use a PaaS. While a bit more than a third of

industrial developers currently use a PaaS in their development, another 47% are evaluating or planning to

evaluate PaaS, foreshadowing much greater future adoption.

Given, the utility of PaaS for

development and industrial developer’s

embrace of DevOps and similar disciplines,

a more ubiquitous current utilization of

PaaS might have been anticipated.

However, PaaS offerings are still in their

infancy with PaaS solutions tailored

specifically for industrial developers only

relatively recently being introduced.

What are your plans for using or integrating Platform-as-a-Service (PaaS) in your development work?

Valid Percent

Currently using 35.3

Evaluating 35.2

Planning to use/ planning to evaluate 11.6

No plans 17.8


Percent of Programmers




Not surprisingly, managing security had the plurality of responses; security is always a major concern for

developers regardless of segment. However, in the case of Industrial IoT, it has enhanced importance in that

physical things are involved and thus the

hazards and vulnerabilities are increased.

Multiplicity of Device Targets

It’s to be expected that in a study of industrial software developers, a top target for connected devices

would be specialized industrial machine controls, and in this survey the plurality of 38% currently targeted

industrial controls. However, connected home devices and consumer boards were close behind at 33% and 34%

respectively. When we look at plans for the next 18 months, all of the possible targets were within the margin of

error.

This points out the fact that industrial developers are not constrained to industrial machine controls, and

that while those are important device targets in this sector, the activities of these developers are far more wide

ranging than one might expect.

What are the top three challenges for building / writing IOT apps? Valid

Percent

Hard to manage security 37.4

Lack of standards for communication protocols 31.7

Lack of access to specialized target devices / data streams 30.4

Lack of standards for embedded systems (OS) 28.2

Lack of easily available, well documented APIs 26.9

Hard to manage data volumes / velocity 22.9

Lack of understanding of deployment architecture 18.8

Need for specialized domain knowledge 17.2

Lack of understanding on approaches to building IoT Apps. 14.7


The general industrial developer

is also concerned with the lack of

standards for communication protocols

and the lack of specialized target

devices. With the exception of security,

which always tends to be a significant

concern, the top three challenges stem

at least in part from the relative newness

of IoT development.


Perc

ent

of Pro

gra

mm

ers



Conclusion

The nature of industrial development in its current stage provides a great opportunity for the adoption

of a Platform as a Service cloud-based offering that is robust and well focused on industrial applications.

Industrial developers spend too much time on peripheral tasks not related to actual development which could

be offloaded to a PaaS. They target multiple device types, and they work in multiple industries. Security

concerns are a big issue for industrial developers, especially those related to development for the Industrial

Internet of Things.

The need for a PaaS that supplies security, provisioning, systems availability and APIs that are industry

relevant is clear. A cloud-based PaaS that also enhances the coordination of development and operations,

continuous integration and automated test/build will be directly beneficial to these developers’ productivity.

GE’s Predix offering provides for these industrial development needs and comes at the right time in the

evolution of the industrial software development market to make a significant impact.

Methodology and Demographics

To meet the specified goals, 1,243 industrial software developers were surveyed online over the course

of two weeks. This provides a margin of error of 2.7% with a 95% confidence level.

Respondents for this survey were sourced from Evans Data International Panel of Developers and were

required to be writing or planning to write specialized applications for one or more of several industries:

including Healthcare, Energy Management, Oil & Gas, Transportation and Manufacturing.

Industrial developers in this study are full time professional developers and, as their name suggests, they

specialize in writing applications that monitor and diagnose industrial equipment. They are well educated with a

majority holding at least a master’s degree. Seventy-two percent of the developers in this study work in small to

medium sized companies with anywhere up to 1,000 employees Team size tends to be small to medium and

58% of industrial developers work in teams of between 2 and 25 individuals.

They focus on a variety of industries from aerospace to oil and gas to manufacturing. Interestingly,

developers across most industries are working primarily on applications that monitor and diagnose industrial

equipment. Embedded applications for control systems are almost as popular.

The prevalence of monitoring and diagnostic development among the general industrial developer

population suggests that to best serve this community, tools and libraries that support and facilitate

development of this type are essential.

Evans Data Corp. (EDC) has made every effort to produce the highest quality research product in this effort. The

customer understands that EDC uses those statistical and data gathering techniques which, in its opinion, are the most accurate

possible. However, inherent in any statistical study is a possibility of error, which must be taken into account in evaluating the

results. Evaluations and interpretations of statistical research findings and decisions based on them are solely the responsibility of

the customer and not EDC. The conclusions, summaries and interpretations provided by EDC are based strictly on the analysis of

the data gathered, and are not to be construed as recommendations; therefore EDC neither warrants their viability or accuracy

nor assumes responsibility for the success or failure of any customer actions subsequently taken.

About Evans Data Corp

Founded in 1998, Evans Data Corporation was created to fill the demand for market research, market intel-

ligence, and strategic planning in the software development industry. Since then we have become the industry

leader in market intelligence focused on all areas of development from software to hardware to mobility.

At EDC we have in-depth and focused experience working with high-tech professionals, and we specialize

in conducting market research in the IT and development community. We are experts in analyzing technology

trends and attitudes. We know how to speak the language and ask the right questions and more importantly, we

can understand the answers.

Evans Data Corp

340 Soquel Avenue

Santa Cruz, CA 95062

Phone: 800-831-3080

www.evansdata.com

Documents

4UBUF PG *OEVTUSJBM *OUFSOFU QQMJDBUJPO …predix-io-assets.s3.amazonaws.com/2016 State of Industrial Internet... · almost two-thirds of indus-trial developers apportion more than

4UBUF PG OEVTUSJBM OUFSOFU QQMJDBUJPO …predix-io-assets.s3.amazonaws.com/2016 State of Industrial Internet... · almost two-thirds of indus-trial developers apportion more than