Architectural approaches for success

In this section, we will look at the need for an Industrial Internet-centric architectural approach to be successful in delivering the business outcomes. Just as civil engineers and building architects use blueprints to incorporate best practices in their work in a reusable way, reference architectures for IT systems have been extensively defined and used to prevent the reinvention of the wheel again and again.

Here, we will focus on reference architectures for IoT and more specifically on emerging reference architectures for the Industrial Internet and IIoT projects. To fully understand such reference architecture, a familiarity with system design principles, enterprise architecture, security frameworks, and networking architecture will be highly useful.

Reference architectures for the Industrial Internet

Reference architectures for the Industrial Internet can be very useful in facilitating the communication between the architects and the stakeholders in industrial manufacturing domains, including plant managers, field engineering managers, service professionals, business managers, and others. The solutions tend to address very specific business problems such as determining fuel efficiency and when engine maintenance is required. IT-centric architecture frameworks are less useful for understanding how the convergence of OT and IT will provide a means to achieve the business outcomes expected from the Industrial Internet solutions. However, there is a need for the reusability of this underlying IT architecture to scale the lessons that are being learned broadly.

Architects refer to the reference architecture and use it as a template as they capture the requirements. They design the specific implementation of the architecture and can convey a consistent understanding to internal and external stakeholders. Thus, interoperability, security, and other requirements are addressed upfront and do not become an afterthought.

Reference architectures lay the foundation for best practices and the reuse of the architectural patterns. As per the US Treasury Architecture Development Guidance (TADG) publication (http://pubs.opengroup.org/architecture/togaf8-doc/arch/chap28.html), the definition of a pattern is an idea that has been useful in one practical context and will probably be useful in others. The structure of the pattern can include some of the following elements:

• Name: Easy-to-remember nomenclature
• Problem statement: Description of the challenge to solve
• Context: The current state where the pattern could be applied
• Forces: The internal and external drivers and constraints; this could include the regulatory landscape as well as the societal implications
• Solution: The details of how to solve the problem at hand
• Resulting context: The outcomes and the trade-offs
• Examples: Sample applications
• Rationale: The why and the detailed explanations
• Related patterns: How this pattern is similar or related to others
• Known uses: Where this pattern is in use

Throughout this book, we will see the evolution of and use of architecture patterns in the context of the Industrial Internet. For example, there are different patterns for gateways and edge architecture. New cloud-based architecture patterns continue to be introduced.

The Industrial Internet Consortium (IIC) has recognized the need for the reference architecture and has published the Industrial Internet Reference Architecture (IIRA). This three-tier architecture provides different view points targeted at the different stakeholders. IIC defines the reference architecture as the output of the application of architecture principles to a class of systems. This is used to provide guidance as the architects analyze and solve the common architectural concerns. The resulting IIRA then provides a template for use in the concrete architecture of Industrial Internet systems.

IIoT projects and architecture solutions can be extremely complex. A proven approach to solving complex problem design is to decompose it into its subsystems. So, to further accelerate the adoption of the Industrial Internet and enable delivery of the desired business outcomes, similar analytics, security, and connectivity frameworks are provided by IIC.

The multi-tier IIoT architecture

Next, we will take a look at the tiers of the architecture and how they interact to produce the desired system behaviors. In subsequent chapters of this book, we will provide guidance on how to simplify the design and analysis of the subsystems and foster their reusability.

The most commonly used reference architectures for the Industrial Internet and IIoT have three-tiers: Edge tier, Platform tier, and Enterprise tier. The commonly used definitions of the three tiers are as follows:

• Edge tier: The Edge tier collects data from the deployed machines (the sources of data) using various connection types. The architectural concerns for the Edge tier can include the nature of sensors and the machines or devices where data is being collected from, their location, governance scope, and the type of network connection, as well as the storage, transmission, and the computing needs for the collected data.
• Platform tier: The Platform tier receives, processes, and forwards data and control commands from the Edge tier to the Enterprise tier and vice versa. It can provide structures for data ingestion, data stores, and asset metadata, and can store configuration data and provide non-domain-specific services such as data aggregation and analytics.
• Enterprise tier: The Enterprise tier can implement domain-specific applications, decision support, and business intelligence systems, and provide user interfaces to human consumers of the information.

Let's take a quick glance at the following diagram depicting the tiers mentioned earlier:

The providers for Industrial Internet platforms and solutions decide what functionality to provide and which components to configure in each of the tiers. General Electric often uses the terminology get connected, get insight, and get optimized, which requires all the three tiers to fully realize outcomes from Industrial Internet. A more detailed review of typical IIoT platform strategies and solutions will be covered in subsequent chapters.

A security framework for the Industrial Internet

Industrial accidents can cause devastating damage (as witnessed at Fukushima, Chernobyl, and Bhopal) with large-scale damage to the environment, injury, or the loss of human life. As we enable software-based systems to increasingly interact with operations of critical infrastructure, there is an increasing need for robust security frameworks for the Industrial Internet.

The IIC has defined an Industrial Internet Security Framework (IISF). The IISF is a collective work product of security experts from companies such as ABB, GE, Intel, RTI, as well as academicians from JHU and UPenn. It was reviewed by professionals from Oracle, Microsoft, and IBM to name a few. Such cross-company initiatives prove that security frameworks and best practices cannot be over-emphasized for Industrial Internet.

The purpose of IISF is to provide a point of view on the security-related architectures, designs, and technologies, and identify procedures relevant to trustworthy Industrial Internet systems. IISF describes the security characteristics, technologies, and techniques needed to address security concerns and gain the assurance that system trust worthiness is achieved.

Apart from the traditional concerns of hacking and theft of information, resiliency is a key concern for Industrial Internet systems. IIC defines resilience as the condition of the system that allows it to be able to avoid, absorb, and/or manage dynamic adversarial conditions while completing assigned mission(s), and to reconstitute operational capabilities after casualties.

For example, a smart thermostat controlling the HVAC system in a building could receive a command to raise the building's temperature by 50 degrees Fahrenheit in the next hour. It is known that the building was operating in a normal temperature range for the human occupants in the building. The resilience built into the system would prevent a sudden rise in temperature and would either create an alarm for this command or include a reliable human in the loop of the decision making before acting.

This kind of system is called a Cyber-Physical System (CPS). According to the National Science Foundation (NSF), CPSs are engineered systems that are built from, and depend upon, the seamless integration of computational algorithms and physical components (https://nsf.gov/funding/pgm_summ.jsp?pims_id=503286).

Future research and advances in CPS will enable capability, adaptability, scalability, resiliency, safety, security, and usability that can be used for the benefit of Industrial Internet systems. CPS technology will drive innovation and competition in the industrial sectors such as agriculture, energy, transportation, building design and automation, healthcare, and manufacturing.

A connectivity framework for the Industrial Internet

Connectivity is at the heart of making IIoT projects functional. Interoperability among the tiers and to devices must be planned for.

IIC released the Industrial Internet Connectivity Framework (IICF) to deal with such architectural considerations; this is a result of contributions from professionals in several large companies, including Cisco, Ericsson, Nokia, RTI, GE, Samsung, AT&T, and SAP. The IICF links the different elements of the rich and diverse landscape of the Industrial Internet and defines an open connectivity reference architecture. IICF enables architects to evaluate and determine the suitability of a connectivity technology for their systems and solutions.

The IICF answers commonly asked by architects about connectivity in the different layers and their functions for Industrial Internet systems. It describes the architectural characteristics and design trade-offs at each layer, commonly available industry standards for these layers, and the categorization and evaluation of the relevant connectivity technology. As we dig deeper into the connectivity issues for Industrial Internet systems, in upcoming chapters, you will learn about connectivity frameworks often used in manufacturing such as Open Platform Communications Unified Architecture (OPC-UA), developed by the OPC Foundation. OPC-UA is a machine-to-machine (m2m) communication protocol commonly used for industrial automation.

The industrial data analytics framework

The industrial data analytics framework describes dig data analytics management systems on Industrial Internet systems data, which often take the form of the following data:

• Relational data: This format is best suited for metadata of assets and things, and as it captures the system configurations and relations to enterprise data systems. Commonly used relational database systems are Oracle, Microsoft SQL Server, IBM DB2, MySQL, and PostgreSQL.
• Time series data: This is a series of discrete data points in time order, often equally spaced in time. For industrial assets and sensors, this may be the bulk of the data. Such data is often stored in historian software that records the historical information and trends about industrial processes. NoSQL databases are also used to manage this type of data.
• Object related data: This form of bulk object storage is best suited for images, blobs, and other unstructured data. Examples of this type of storage are Amazon S3, Microsoft Azure blob storage, and Scality that can be deployed on-premise.

To run industrial analytics on such a variety of data formats, real-time and batch capabilities are required. The ability to orchestrate multiple analytics workflows is also required.

The stakeholders for analytics can be data scientists, analytics developers, architects, as well as subject matter experts (SMEs). The following diagram illustrates the typical life cycle of the development of industrial analytics:

This is an iterative process and suitable for agile development. An important characteristic of the industrial analytics is the ability to not only pull the aggregated and summary data to the analytics but also to be able to push down the analytics to near real-time data feeds. This is due to the extremely large data volumes that devices transmit and the frequent nature of these transmissions.

In subsequent chapters, we will talk about such near real-time analytics technologies and discuss the emergence of the NoSQL database and Hadoop-based data management solutions fundamental to solving these problems. Architects of Industrial Internet solutions must embrace skills in industrial analytics and new data paradigms to be able to design effective solutions.

Cloud and user experience considerations

The frameworks we've described provide some of the background material you will need. However, other areas are less well covered because they are relatively new (cloud computing) or not typically in the domain of architects (user experience).

Architects are typically very comfortable defining on-premises systems and translating that knowledge into public cloud concepts. IaaS is a form of cloud computing that provides virtualized computing resources for enterprise and Industrial Internet systems in the form of operating systems, servers, storage, and, networking. They can also choose PaaS which also delivers data management systems, tools, and the management of those components, or SaaS that provide an even more complete solution. We will further discuss the business and technical trade-offs of each in the next couple of chapters.

That said, architects must make key design decisions that span the on-premises and Public Cloud paradigms. They must consider where the data is stored and who it belongs to. The compliance and regulatory landscape often becomes a key consideration for the architects. As with any solution, they must consider who has access to the data and under what context. We'll provide much more guidance here later in the book.

As data is collected and analyzed to turn insights into action, the user experience, or UX, assumes importance. It is important to remember that UX Design refers to the User Experience Design, while the more understood UI Design stands for User Interface Design. An industrial worker on the factory floor or a field service technician working on overhead power lines has a very different expectations when interacting with the environment and the device they use to deliver actionable tasks.

Business strategy framework for the Industrial Internet

While architects often confine themselves to technology challenges, in the realm of the Industrial Internet, business considerations go hand in hand. IIC provides a business strategy framework that illustrates the major areas that should be of interest:

Industry analysts agree that the IIoT is experiencing explosive growth, and emerging leaders in companies, such as the CDOs, are being tasked with driving strategies. Architects must sharpen their business acumen and have an opportunity to groom up for future digital leadership roles.

Areas where architects can broaden their contributions to their companies will include, but not be limited to, the following things:

• Understand the competitive landscape and help define their company's role
• Understand new market dynamics and pressures introduced by IIoT
• Understand new business models, the value chain, and partnerships/alliances and continuous reevaluation of the same within their company
• Evaluate the societal impact of the Industrial Internet