Harnessing the Power of Big Data for Digital Manufacturing
In our introductory article about the Birth of Digital Manufacturing, we provided a glimpse into the future – our view of what the world of digital manufacturing might look like ten years from now. In that view of a potential future, fictional CEO Connor Smith described Entity-X’s remarkable transformation from a small fabrication plant to a thriving multi-billion-dollar company producing smart connected products for businesses and consumers. Key to that transformation was Entity-X’s smart use of data, which fundamentally impacted all phases of the product’s entire lifecycle, from ideation, to design, to manufacturing, to in-field support.
Our story is not just fiction. These changes are already happening – companies are already taking advantage of the confluence of emerging new technologies such as IoT, Cloud computing and big data to gain new insights and provide compelling new products and services. Of all these new technologies, big data may prove to be the one with the broadest impact on the largest number of companies. More than likely, your company will face not only exciting opportunities for innovation, but also terrifying threats to your business.
So, as a leader of engineering, manufacturing or operations, what should you do?
What is Big Data?
Big data is any data that’s too big to be handled by traditional database technologies, where the bigness is measured along three primary dimensions:
- Volume – The shear amount of data to be stored and processed
- Velocity – The required speed of data flow, including input/output, processing, and storage
- Variety – The variability (and unpredictability) of the kinds of data to be stored and processed
For example, consider the GE Durathon battery plant that collects data from more than 10,000 sensors in real time. This system has high Velocity due to the data being generated and processed at a rapid rate, high Volume due to massive amounts of data accumulating over time, and high Variety due to different sensors generating many different data formats.
Traditional database technologies simply can’t handle the extremes along any of these three dimensions, so entirely new types of technologies have emerged over roughly the last 15 years, with more emerging rapidly. And there is no one-size-fits-all technology – different kinds of big data problems require different kinds of solutions, and in most cases a mixture of solutions is required. For example, maybe the battery plant has different databases for real-time control of operations vs. batch processing of analytics.
As the Volume, Velocity and Variety of data continues to explode, so does the array of big data technologies. In fact, according to a recent IDC forecast, the Big Data technology and services market will grow at a 26.4% compound annual growth rate to $41.5 billion through 2018, or about six times the growth rate of the overall IT market.
What does all of this mean to you? As you work to define and execute your big data strategies and plans, you will likely continue to be faced with a rapidly changing technology landscape, with new problems to be solved and new technologies presenting new opportunities for innovation.
Of course you’ve been hearing all the buzz about big data for several years now, but do you really know what it means, what kinds of problems it can solve, and what opportunities it presents to your company? Do you have a deep understanding of what your company and other companies are already doing with Big Data? These are important questions to answer, and they are moving targets, so you need to get educated and stay educated.
Establish a Cross-functional Big Data Team
You next step should be to ensure that your company is taking a holistic view of big data, and to consolidate any existing silos of big data activities and expertise. Don’t limit yourself to thinking that big data is only about analytics or only for sales and marketing, and don’t leave it to IT to figure out your big data strategy all by themselves. All aspects of your business will be affected, so you need a cross-functional team effort to develop and execute a comprehensive business strategy and plan regarding your data sets. This team needs to range from strategic business thinkers to PHD data scientists, and should include translators – people with expertise in both analytics and business who can bridge the communication gap between data scientists and business leaders. Beware that people with deep expertise in analytics are currently in short supply and command premium salaries.
Consider that many companies are already creating big data business units and corresponding executive leadership positions, such as Ford Motor Company’s recently appointed Chief Data and Analytics Officer. In fact, Gartner predicts that 25% of organizations will have a Chief Data Officer (CDO) by 2017, and they argue that the CDO should be a peer of the CIO, as summed up by this statement: “Data management is on the same level, organizationally, as financial management and technology management — and just as important.”
Establish a Big Data Business Strategy and Plan
The first job of your cross-functional team should be to start formulating a big data strategy and plan as a key component of your company’s overall strategies and plans. But don’t let your current business strategies limit your thinking – you may discover new opportunities and new business models, just like Entity-X did in our story.
Your goal should be value creation through both topline and bottom line growth. For example, cost cutting measures might present themselves in the form of modified manufacturing processes, and you may realize myriad new product ideas.
Some companies have adopted a strategy of “let’s gather all the data we can and then we’ll figure out how to analyze it for actionable insights later.” That’s not a strategy. That’s backwards. Technologies like Hadoop and Cloud computing make it technically and economically feasible to collect massive piles of unstructured data and then figure out how to process it later. But first you should have clear objectives, clear understanding of risks, and a much broader view of big data than just analytics.
Pay Special Attention to Privacy, Security, and Data Ownership
Before you decide what data to collect and why, you need to have a clear understanding of and well-reasoned policies for the critical issues of privacy, security and data ownership. These topics should be outlined in your business strategy. They are big topics, but we can offer a few key recommendations:
- Do the Right Thing – Obviously you have to comply with all applicable laws and regulations, but that’s not enough to eliminate risks. The laws are still evolving, and customer trust is not easily gained or kept. So go above and beyond what’s required, and be judicious in your choices about what data to collect and why. Be completely open with your customers and treat them how you know they want to be treated – the golden rule.
- Provide Clear Value to Customers – Don’t just collect information that is self-serving. Make sure it has obvious value to your customers too. For example, Entity-X’s Cloud Baby product collected a lot of private and potentially sensitive data about the baby, but that data, as a result of added services and insights, was valuable to the customer too.
- Consider How to Monetize the Data – Who owns the data that you collect – you or your customers? If you have consent from your customers to share their data, it could open up new opportunities. For example, imagine how valuable the Cloud Baby data could be for medical research, especially if correlated with health data provided by each baby’s pediatrician. Imagine if heart rate, breathing and video images of sleeping patterns could be used to help predict and prevent SIDS (Sudden Infant Death Syndrome, the sudden death of babies during sleep with no warning signs or clear reason).
- Understand the Technical and Human Challenges – In our story about Entity-X, we talked about the huge value of democratizing the data – allowing everyone in the organization to see and make use of data. But what if some of that data is sensitive? Because big data tends to be unstructured, it can be technically difficult to provide granular security (to allow only certain people to see certain kinds of data). And more importantly, human factors tend to be the biggest challenges for enforcing privacy and security.
Fix Your Small Data Problems First
Don’t make the mistake of rushing to do what’s new and cool at the expense of excellence in the fundamentals such as PLM, ERP, MES and CRM. All of your big data knowledge and insights will be useless if you can’t translate them into actions. And your newfound big data insights will actually increase the demands on your traditional business systems, not reduce them. Consider what CEO Connor Smith said: “…with the added requirement to constantly update existing products in the field and churn out entirely new ones, PLM needed to integrate into a larger innovation platform covering a broader scope of data,” and “Now we had PLM extending out to the product’s life in the marketplace…”
Don’t Underestimate the Technology
If you think of big data technology as “Hadoop and analytics,” then your view is too narrow. There are many different classes of big data problems that require different kinds of technology solutions. For example, imagine if the Cloud Baby product had a web application called Babybook that allowed parents to create video/photo albums, track baby’s health statistics, and interact with their pediatrician to ensure baby’s health. And imagine that millions of parents are using this application at once. That’s not the domain of Hadoop, which is better suited for batch processing and analytics than real time operations. This would be the domain of some kind of NoSQL database, of which there are many choices. And the list of big data technology choices is exploding. The research firm IDC forecasts that the market for big data technology and services will grow at 21.7% CAGR, or about six times the growth rate of the overall information technology market. So you will need to do your homework and stay on your toes.
The wide array of technology choices may seem daunting. Here are some suggestions to help you get started:
- Start Small. Look for early wins that you can showcase and promote within the organization
- Use strategy to guide your technology objectives. Ensure senior executives in charge of important initiatives are engaged. Avoid unbridled technology quests.
- Get some help at first. Consider hiring some outside expertise to help you get started, and build your internal expertise over time. The supply of experienced data scientists and practitioners is scarce.
- Think carefully about what to build vs. buy. Right now you might have to build a lot yourself, but that is changing rapidly, which could actually be an advantage to later adopters who buy instead of build.
- Use the Cloud. Don’t even think about doing all of this on premise. In the cloud, the security is better, the economics are better, and it’s much faster and easier to get started and to experiment. Consider packaged services that include both infrastructure and software in a PaaS or SaaS model, such as Amazon’s AWS IoT.
Think Big, but Start Small and Start Now!
All of the steps mentioned above will take time to accomplish and will be ongoing. These steps are important – you need a business strategy and plan to guide your efforts, and you need to avoid letting big data efforts drain resources away from critical small data efforts. But don’t let that stop you from starting now. Start small with a new product idea, just like Entity-X did with Cloud Baby. Research, experiment, prototype. And most of all, innovate!
A Look Ahead: An Interview from the Future on the Birth of Digital Manufacturing
Volume, Velocity, Variety: What You Need to Know About Big Data
Gartner Says CIOs and CDOs Must 'Digitally Remaster' Their Organizations: Gartner Predicts 25 Percent of Organizations Will Have a Chief Data Officer by 2017
IDC Research: Forecast of Big Data Technology Growth
In praise of “light quants” and “analytical translators”
Big data: The next frontier for innovation, competition, and productivity
Sodium Battery Technology Improves Performance and Safety
A Look Ahead: An Interview from the Future on the Birth of Digital Manufacturing
Two Ways IoT is Disrupting (and Helping) the Manufacturing Process