This is a Wikibon Voice of the Community Report, sponsored by Dell EMC. Voice of the Community posts are identified paid posts that appear on all pages of SiliconANGLE.com, supporting editorial efforts.

Premise

IIoT&P edge computing is gaining traction in business. New hyperconverged infrastructure technologies can be employed to add scale and flexibility to most complex edge environments.

Introduction

The Industrial Internet of Things and People, or IIoT&P, is reshaping business. Once the province of supervisory control and data acquisition, known as SCADA, in discreet and process manufacturing verticals, businesses of all sizes and shapes are exploring use cases for combining IIoT&P sensor, actuator, analytic and cloud technologies into differentiated business systems that can reduce costs, improve operations and enhance customer experience.

However, as IT leaders plow the application greenfield for IIoT&P, they are encountering physical, infrastructure, cost and organizational realities that demand considerate, architectural treatment. A new set of technology choices and disciplines – edge computing – is emerging to address these realities.

But what is edge computing? Is it just general-purpose SCADA? A last-ditch attempt by information technology traditionalists to hold back cloud advances into businesses? As an IT professional, is there something deeper that you need to understand to help you choose outcomes, guide strategy and lead changes related to edge computing?

To answer these and other questions, Wikibon convened a CrowdChat to discuss the impacts of industrial IoT&P on technology architecture. For an hour in late January 2017, 60 industry experts who are part of the Wikibon community met online to discuss this topic. The conversation comprised over 200 collaborative observations that generated nearly 600,000 customer impressions. Among the key findings from the collaboration:

• Most businesses will be strongly impacted by IIoT&P.
• Edge IIoT&P requirements will have a dominant effect on technology architecture overall.
• Edge technology selection will be driven by data physics, regulation, security and cost realities.
• Managing IIoT&P will force a reorganization of IT, OT and technology sourcing strategies.

*******************************************

What Is A CrowdChat?

CrowdChat is a community engagement tool used by Wikibon to research innovation. A CrowdChat brings together – online – experts in a domain to discuss complex technology, social and business issues. Wikibon posts questions to these experts, which catalyzes a bloom of conversational interactions about the subject. Wikibon analysts then combine these interactions with other research sources to develop the findings that we publish in a Voice of the Community research paper.

*******************************************

The impact of IIoT&P on business

IIoT&P forces a rethinking of the relationship between business and information technology. For most of the first 50 years or so of the IT industry, business applied relatively known processes to relatively unknown technology. For example, accounting processes were codified as accounting software packages on rapidly evolving hardware technologies. While the characteristics of the application processes evolved, often in response to technology possibilities, generally these systems used highly stylized data and process structures that were defined by conventions and rules that could be, and were, established by human institutions.

IIoT&P flips this relationship (see figure 1). Today, through IIoT&P, businesses are trying to use relatively well known cloud, sensor, and robotic technologies to execute unknown processes – like customer journeys or the paths taken by autonomous vehicles. IIoT&P systems must address real world activities and analog data types, in contrast to the stylized, “natively digital” process of past IT systems. Successfully addressing this rich, and richly complex, world of IIoT&P requires technology professionals to understand that:

• IIoT&P is an information transducer. The significant majority of business activities take place in the analog data world. People interacting with people; people interacting with things; even many electromechanical devices that interact with each other do so through analog signals. A seminal feature of IIoT&P is that it transduces analog and digital data, both turning analog data into digital signals and digital results back into real-world events through the systems of agency that perform real world work and provide real world services on behalf of firms and people. Crucial to this flow are analytic models, which process data signals and generate the instructions that tell actuators what to do.
• IIoT&P use cases are vast and will be new sources of business innovation. How extensive will be IIoT&P use cases? Can you think of a part of your business that changes in response to independent customer behavior, environmental conditions, or sensor-based automation? If the answer is yes, you’ve identified a possible IIoT&P use case. While IIoT&P will remain crucial to production control, the basic concept of IIoT&P as data transducer will mean that IIoT&P concepts will increasingly feature in mobile, customer engagement and business automation applications. For example, Schindler, the 143-year-old elevator, escalator and moving walkway company is rethinking its products as services, exploring how it can use IIoT&P to not only improve energy efficiency and maintenance scheduling, but also to improve the experience of “passengers” as they move through industrial and commercial spaces.
• All businesses will be affected. IIoT&P concepts and technologies are a centerpiece of digital business transformation. Why? Because at its core, digital business transformation is about how businesses can better create, manage, and apply digital assets. Any business that seeks to use digital technology to better anticipate customer decisions, automatically alter products and services in response to environmental or other contingencies, or incorporate robotics is going to have to master some aspect of IIoT&P.

Figure 1. IIoT&P transduces rea-world “analog” data and digital-world “digital data.”

*******************************************

What does the Wikibon community say about IIoT&P business impacts?

• “It’s hard to think of a vertical that will not be revolutionized by IIoT&P. As the cost of computing approaches zero, IIoT&P will invisibly spread into everything.”
• “Health won’t be able to cope with the exploding costs of traditional hospital settings. IIoT&P in home-based care is an important trend.”
• “Focus on the business case first, and then worry about what to instrument.”

*******************************************

Welcome to the ‘edge’ of technology architecture

IIoT&P’s impact on business will be significant; according to the Wikibon community, its impact on technology architecture will be even greater (see Figure 2). IIoT&P systems frequently must distribute data, processing, and control out to where the action is. This is what is meant by “IIoT&P edge”: The resources required to process and perform IIoT&P events typically must be proximate to sites where the events take place. This is in direct opposition to the notion that all processing will move to central, cloud locations. While important business computing workloads will migrate to the cloud over time, edge computing will countervail, driving greater distribution of computing resources. Why? Because:

• The edge involves unprecedented architectural constraints. Applications at the edge often demand extremely low latency, capture highly sensitive data, operate in environmentally challenging and remote locations, and require extreme security. For example, the difference between safe operation and a blow out at a well head can be a few milliseconds. Data collected from sensors that indicate a potential disaster can’t be shipped hundreds of miles to an analytics system running in a central cloud location if the roundtrip time to get a decision exceeds the time required to act. As businesses further embed digital technologies for automation or enhancement, the integration of edge computing with other forms of computing will require complex architectural decisions. Because edge will be so impactful to a business’s safety, efficiency and value delivery, edge architecture considerations frequently will dictate overall IT architectural choices.
• Data movement costs are significant and likely to rise. However, even when an edge implementation features more normal constraints, the costs of data movement from edge to cloud can be prohibitive, forcing architectural designs to keep data at the edge. For example, Wikibon studied the costs of edge processing for a relatively modest-sized wind farm (see Figure 3). Our research showed that the data communications costs of moving sensor and control data back and forth from edge-to-cloud overwhelmed all other costs. Modest-sized edge sites that require hundreds or thousands of sensors operating at routine power levels can still throw off enough data to make cloud integration impractical for most data flows.
• Edge technology options are packaged and priced right. It used to be that robust, scalable hardware technologies couldn’t operate without significant administrative attention and purified environmental. That no longer is the case. The computing technologies that have taken over the data center – virtualized and software-defined, rack-based CPU, flash storage and integrated networking platforms – are being packaged to support edge computing needs at almost any scale and environmental conditions. For example, Dell EMC’s VxRail appliance can be pre-configured to support a range of analytics, automation and lights-out operation requirements. These hyper-converged platforms enable data center performance and administrative control at the edge, limiting the need for trading off processing quality, cost and IIoT&P fidelity. Wikibon calls them “edge centers.”

Figure 2. How will IIoT&P edge shape your technology architecture? (Source: January 2017 CrowdChat on edge centers)

Figure 3. Data movement costs can make public cloud options impractical
(Source: Wikibon, “The Vital Role of Edge Computing for IoT: 2016 Update, 8 November 2016)

*******************************************

What does the Wikibon community say about edge impacts on technology architecture?

• “The volume of machine data will significantly restrict the ability to transmit to data centers.”
• “Most telemetry today runs at 1Hz or lower frequency. Moving to advanced analytic applications that require machine data at 10s-to-100s Hz will certainly be cost-prohibitive to centralize.”
• “For a simple sensor module (~15 sensors) it cost about $800 a month just to move and ingest the data centrally. That will hurt when a business turns on a fleet of sensors.”

*******************************************

The factors driving edge technology selection

Edge computing sites often comprise sensors and devices from dozens of operational technology (OT) vendors, each utilizing different interface specifications, communications protocols, analytic models, and control mechanisms. Gateways have been used to mediate these differences, translating protocols and managing data flows within the edge site and from the site back to central locations. However, gateways typically provide limited capacity for general purpose computing, complex analytics processing or supporting software-defined resources. Moreover, software-defined edge gateways are emerging from companies like Microsoft, providing additional incentives to consider more powerful hyper-converged infrastructure (HCI) platforms that can concurrently run a range of gateway, sensor stream, analytic and management workloads.

During the CrowdChat, we asked the Wikibon community about which criteria should guide technology selections for edge platforms. Ruggedizing always is an issue, but an increasing range of platform options can meet the stringent power, environmental and administrative requirements of edge. The community was not definitive – the conversation didn’t reveal clear demarcations for gateways or HCI systems – but did build consensus that three criteria will drive platform technology selection:

• How real time are the requirements? The dominant consideration at the edge is latency: Roundtrip times, including distance (the speed of light) and software execution times (path lengths) for data cannot exceed the time required for action. In many edge cases, the latency tolerance is milliseconds. The biggest contributor to latency is the distance a message must travel over a network, but other factors, like memory constraints or I/O latency, must be considered. Robust, hyper-converged platforms that can support deterministic processing at scale, exploit flexible and fast flash storage, and can process the complex analytic models being deployed to digitally represent “things” (GE Digital calls these models “digital twins”) will be favored for an expanding variety of use cases. For example, an oil field edge may combine “near-edge” machines to process micro adjustments on a drill bit with “mid-edge” HCI systems that apply data streams for multiple rigs against analytic models intended to optimize operations across the entire field.
• How much data will be flowing at the edge? A second consideration is the volume, complexity, and security requirements of data flow at the edge. The processing requirements for any individual data stream – for example, from a 2Hz sensor to a time-series database – may be modest. However, the practical factors of latency and data communications costs mean that as much as 90 percent of edge data will remain within the edge domain. As more sensors are embedded into devices and the models that make up device digital twins gain complexity, processing requirements can rapidly scale. Moreover, as IIoT&P use cases advance and gain potency, today’s optimized platform becomes tomorrow’s boat anchor. Especially for mid-edge requirements, the community was clear: HCI solutions such as VxRail appliances provide the power and flexibility to handle an increasing range of current and future edge workloads.
• How entrenched are operational technologies? IIoT&P didn’t just pop out of thin air a few years ago. Before IIoT&P, there was SCADA in manufacturing and related verticals, which captured tens of billions of dollars of annual technology investment. In many situations, these investments are like legacy technologies: hardware vendor-specific, proprietary, relatively inflexible, expensive to integrate, difficult to enhance, and understood by a small cadre of personnel. Most edge investments will be “brownfield,” taking the form of enhancements to existing devices. As new edge investments are made, pressure to create “software-defined” versions of legacy IIoT&P elements will increase. Where will these software-defined devices run? Most likely not on gateways, but on higher-performing HCI edge platforms. This won’t happen overnight. However, the community expressed belief that HCI platforms provide the most likely means to accommodate the eventual migration of specialized hardware devices into software-defined devices.

*******************************************

What does the Wikibon community say about edge technology selection?

• “At the moment, due to limitations on gateway scalability, a lot of data at the edge is dumped before its true value can be assessed.”
• “90 percent of IIoT&P opportunity is brownfield . . . Lot of old equipment and entrenched administrative procedures in place.”
• “Planning for headroom at the edge [is key]. There is uncertainty regarding how much analytics should happen at the edge. You don’t want to have to roll a truck in within two years to deploy more computing.”

*******************************************

Institutionalizing IIoT&P success

The edge center concept presumes that business will want to combine the disciplines of IT and OT pros at the edge and will need platforms that can accommodate and integrate the requirements of both. And make no mistake, for practical and political reasons, both sides must be addressed.

IT pros bring the tradition of rapid technology innovation and integration with other business systems to the table (see Figure 4). OT pros bring use case knowledge and edge engineering experience. Historically, IT and OT haven’t mixed well, but that must change if each side is to benefit from the strengths that the other brings to edge opportunities. Ultimately, edge technologies will evolve to cost-effectively handle almost any class of physically-possible edge need, including device features, model fidelity, local computing scale, business application integration, and cloud integration. However, just as political imbroglios between IT groups and telecommunications groups forestalled advances in IP-based, software-defined infrastructure, so will tensions between IT and OT.

How will it play out? As technology professionals have learned over the years, economics generally ensures that proprietary hardware gets turned into software or software services. As HCI and other IT-oriented technologies diffuse into the IIoT&P edge, the processing foundation gets put in place to run software-defined edges.

Figure 4. Knitting IT and OT together

*******************************************

What does the Wikibon community say about IT and OT working together?

• “The first thing we look at is operational culture. The technology must integrate with operations or it won’t stick.”
• “Standards such as ANSI/ISA-95, IEC 61131-3, IEC-62541 and the IEEE 1451 series bridge the gap between IT and OT.”
• “We see this being most successful when a ‘head of digital’ has IT and OT in their purview.”

*******************************************

Action item

As edge computing gains business visibility, old rules-of-thumb – and some newer cloud-oriented assumptions – for architecture, system design, sourcing relationships, and technology selection fail. Legacy IIoT&P investments ensure that most edge investments are brownfield. However, technology leaders must presume that software-defined IIoT&P will be combined with advances in sensor technology to dramatically enhance the flexibility and applicability of edge technology to a burgeoning group of use cases. There is no one-size-fits-all edge platform. However, hyper-converged infrastructure options can support performance, administrative simplicity, and future scale needs of more complex edge situations and should be part of any IIoT&P technology palette.

Image: Pexels/Pixabay