Most agree that the first generation of the Industrial Revolution began in the middle of the 18^th century. Let's go back in history to see how it led to the evolution of the IIoT today. The 18^th and 19^th centuries, which experienced the Industrial Revolution saw a transition from the manually intensive manufacturing processes to the mechanization of the manufacturing. This laid the foundation of the modern heavy industries. At the time, most people lived on farms and worked in agriculture. Factories were commonly located close to rivers and streams where they could be powered by water wheels, and there was usually much handwork involved. With the invention of the steam engine, factories could be located elsewhere. The power that was supplied to machinery by steam engines became more predictable, and more processes could be aided by machinery.

Great Britain experienced many technological innovations ranging from the first engine in 1712 by Thomas Newcomen to the steam engine in 1765 by James Watt to the first public railway line in 1825. The Industrial Revolution transformed manufacturing from the home and cottage industry level to a vastly more scalable level. With the introduction of railroads as a transportation alternative to river traffic and horse-driven carriages, faster travel between distant locations was enabled and provided a new means to deliver supplies to factories and products from them. This theme of decoupling the production facilities from the consumers can be seen in today's computing world where remote data centers can be decoupled from the information technology users. Over time, this Industrial Revolution led to a transition from human labor to the use of machines, spread over whole of Europe and to North America, leading to the industrialization of the world. Gradually, this led to consumerism as goods became available, accessible, and affordable.

The increased widespread availability of electricity through power grids and the invention of the assembly line in the first decades of the 1900s introduced the second generation of the Industrial Revolution. Once again, power became more predictable and the amount of space required for power generation in factories was reduced. Production became more optimized through assembly lines, and workers assumed new specialized roles. Motorized vehicles also appeared for the delivery of supplies and transporting finished products, thus enabling more variation in factory locations. We began an age of mass production as well as mass merchandising, which resulted in the creation of many additional, new kinds of job.

In the third generation, business computing was introduced and efficiencies were greatly improved. Mainframe computers became widely available with subsequent pricing adjustments, making them more affordable and hence more widely adopted in the 1960s. Still cheaper minicomputers and then personal computers followed. The Internet was in common usage for networking within companies and across the world by the 1990s.

The Internet evolved from a way for the military to connect and communicate and appeared in universities and then mainstream companies. The mid 1990s saw the transition from the military's Advanced Research Projects Agency Network (ARPANet) to the consumer Internet. In this wave, computers and servers connected across the world and then provided an information super highway for the people. This revolutionized how people interacted with each other and with the businesses leading to the growth of e-commerce and social media. New leaders emerged in this era, starting with Information Technology (IT) system providers, and companies such as Amazon, which started with online sale of books and went onto become a general-purpose e-commerce platform. Likewise, on the human interaction front, emails became mainstream and more interactive and rich multi-media evolved on the web. This led to the rise of Myspace, Facebook, Twitter, and similar social media platforms to essentially connect the people across the world. We refer to this as the consumer Internet.

Computing also became more accessible through improved software development tools and through business applications and tools that provided increasingly more intuitive user interfaces. Some refer to this as the beginning of the information age as line of business users could access and manipulate their data to measure and optimize their business activities.

While the consumer Internet focused on connectivity between businesses, consumers, the IT systems, and the computing devices such as servers, PC's, laptops, and emerging mobile devices, it largely ignored the machines from the Industrial Revolution. This led to a great divide between the machines for industrial operations and the traditional IT systems and set the stage for the fourth wave that we call the Industrial Internet Revolution.

The Industrial Internet can be defined as the connecting of industrial-grade machines and devices to networked computing devices with the goal of collecting the diverse data originating both inside the machine and the surrounding environment and processing or analyzing this data for meaningful outcomes. Such data originates in various forms and is often referred to as big data. The systematic organization and analysis of this data is referred to as big data analytics for industrial outcomes.

The Industrial Internet and IIoT are the industrial flavor of the IoT. While IoT refers to any physical object or thing connected to a network and the Internet, IIoT focuses on scenarios where the connected objects are primarily industrial in nature (such as manufacturing assembly lines, power generation equipment, or mass transportation vehicles). Thus, Industrial Internet is often used interchangeably with IIoT. There are three Ps important to an industry:

The following illustration captures the interaction of these three Ps:

Industrial machines and assets have a long life, especially when compared to many consumer devices. The following table serves as a reference to highlight the difference of scale in usable life comparing various industrial assets to a smartphone:

Due to the long life and cost of ownership of industrial machines, it is important to provide ways to protect the investment in these machines over time. Thus, the optimization of field maintenance services is an integral part of the Industrial Internet. Service execution and service delivery platforms and applications are within the realm of the Industrial Internet architects, and this book will provide coverage to it.

The long life of industrial assets leads to two terms often used in the context of Industrial Internet solutions: greenfield and brownfield applications. A greenfield project refers to a scenario where a company decides to build a new infrastructure since it offers the maximum design flexibility and efficiency to meet a project's needs (an existing infrastructure limits the ability to change by its present design). From the Industrial Internet architecture perspective, the new infrastructure can add sensors to collect relevant data.

Brownfield projects leverage infrastructure that is already in use. The costs of starting up are usually greatly reduced with this approach, but it can be more difficult to modernize the infrastructure and incorporate the addition of sensors. Construction and commissioning times can be minimized using this approach. For Industrial Internet projects, brownfield systems can be retrofitted by adding external sensors to collect data. For example, external acoustic sensors might be added to the body of air compressors in a factory to do the harmonic analysis and determine air leaks in a brownfield project. Air leaks can cause wasted electricity in manufacturing plants where compressors are used to drive several pneumatic tools.

Some of the concepts we associate with the Industrial Internet today began to mature in the last few years. For example, in a world before widely available smart sensors, oil and gas exploration companies brought computers to the exploration sites, processed the data locally in relational databases, and transmitted the processed data and their conclusions back to their headquarters. Some referred to this as early edge computing on the remote computers. The following diagram reflects this type of deployment:

Data warehouses and data marts became common in most businesses. Batch-fed by Online Transaction Processing (OLTP) systems, they became the place to store historical data used to report on current trends and compare current data with past data through business intelligence tools. Of course, this footprint remains common today.

Predictive algorithms were also developed, tested, and deployed with increasing rapidity in certain industries and gained wider adoption over time. Some early use cases included understanding financial market investment strategies and insurance risk, and the prediction of the likely quality of expensive manufacturing processes to better optimize the production.

Each generation became shorter. Moving from the first generation of the Industrial Revolution to the next was a matter of centuries, but the subsequent generations took half the time of the previous change. This implies that future generations may come at a faster pace, and while we are embracing the Industrial Internet, we need to be prepared for the possible next generations as well.

In 2010, the IoT and Industrial Internet became familiar terminology. The World Economic Forum and others declared this to be the next generation of the Industrial Revolution. As in previous generations, several technological advancements came together to enable a new class of solutions and applications, changing business models and capabilities.

Sensors began to be mass produced at ever decreasing costs. As price points, size, weight, and power requirements for sensors decreased, engineers began to create device designs that included them in anticipation of being able to gather useful data on device status as soon as it became feasible. Since smart sensors can also be programmed and updated, they can evolve and become more "intelligent" over time. For example, inclusion of such smart sensors in automobiles led to rapid advancements in the development of autonomous vehicles.

The sensors themselves most often transmit semi-structured data in a streaming fashion. Coincidentally, analyzing mass quantities of semi-structured data became possible a decade earlier through development of NoSQL data engines (and Hadoop specifically) to solve the problems of Internet search optimizations and recommendations. Next generation platforms holding exabytes of data are deployed today by companies in the search engine business.

The development of new and innovative software solutions became more viable for start-ups and smaller organizations as cloud-based platforms became available (mostly eliminating an expensive upfront investment in infrastructure). The cloud also enabled faster time to deployment and elastic scalability that was difficult in classic data centers.

The cost of networking and bandwidth reduced over this time to provide ubiquitous connectivity for the IIoT. Some of the connectivity options and technologies used include Radio-Frequency Identification (RFID), Wi-Fi, Bluetooth Low Energy (BLE), and 2G/3G/4G with 5G on the horizon.

The growing popularity of open source software data management offerings and development tools also helped minimize early costs. Today, as the Industrial Internet has matured, we see many integrated solution footprints and applications that rely on underlying open source components.

The following diagram represents a common architectural pattern often seen in Industrial Internet implementations and is called a Lambda architecture:

The illustration shows streaming data feeds from smart devices. The streaming analytics engine analyzes this feed in real time and will sometimes have machine learning algorithms deployed to process the data. The data lake pictured is most often a Hadoop cluster and is designed to load and store massive amounts of data of all types. As in the previous generation, traditional data warehouses and data marts are batch fed. Business intelligence tools are shown pointed at the data mart, data lake, and streaming analytics engine in our illustration.

We'll describe these components in much more detail in subsequent chapters as we lay out the data and analytics architecture. Obviously, there is also a lot more detail in the information technology platform architecture, which we'll cover as well.

New manufacturing technologies are also now employed in Industrial Internet solutions. Robotics in manufacturing became common in industries where the cost of labor was high, such as in the automotive industry, around the turn of this century. The robotics that were deployed improved the consistency and quality of the products produced and helped to contain costs. The addition of intelligent or smart sensors to newer generations of these devices enabled more functional and flexible capabilities. The wider applicability and growing usage of robotics also led to decreases in their pricing, helping drive further adoption.

Many manufacturers and companies that design products are now experimenting with 3D printing. 3D printers enable the manufacturing of products and components anywhere; such a printer is deployed and accessible via a network. Such technologies are often referred to as additive manufacturing. The ability to print spare parts on demand for industrial machines can have a profound positive impact on the supply chain ecosystem, as the cost of such additive manufacturing continues to decrease.

Artificial intelligence (AI) and machine learning are also enabling more intelligent devices. As devices become self-learning, they can react to changing situations in real time. We'll discuss these topics and other emerging technologies when we explore what is likely to occur in the near and more distant future in the last chapter of this book.

These new capabilities are causing companies to rethink the value of their data and the kinds of businesses they are competing in. Many are facing new and non-traditional competition from other industries and are evaluating digital transformation strategies that sometimes include new strategies for monetizing their data assets. Some are becoming data aggregators, selling data to other companies and subscribers that find it useful.

The following diagram summarizes the four generations of the Industrial Revolution we described:

As always seems to happen when a new generation begins, there are some holdover problems from the old generation as well as problems introduced by the new architecture. One carryover from the previous generation is the need for projects to be driven by line of business requirements, not by IT. As it was earlier, projects will usually stall when IT-initiated proof of concepts do not really solve problems that the business needs and wants to address.

In Industrial Internet projects, architects and IT must also sometimes work with engineering designers who are specifying the types and locations of sensors in devices to assure that data needed for the proposed solution can be gathered. Similarly, these teams need to work together regarding networking requirements given the amount of data that might be transmitted. Continuous data gathering from equipment operated in industrial settings is key to enabling maintenance and field services-related solutions.

The mixture of semi-structured and unstructured data and the variety of data management solutions needed introduce complexity and the need for new skill sets that an organization might not possess and face difficulty in finding. Further adding to the complexity is the rate at which data is transferred over networks arriving in the data management engines and the data volumes that must be managed in them.

Of course, device and data security must be maintained throughout the ecosystem. Software and firmware updates that are pushed to intelligent sensors and devices must be secure and successful, or denied. Data transmitted to cloud-based solutions must meet or exceed industry-relevant certifications and country data sovereignty and privacy laws.

External threats can exploit vulnerabilities in under-protected Industrial Internet systems and thereby cause harm to the organization owning the assets and the associated business processes. Such concerns led to an increased focus on solving these security risks and adopting the emerging standards.