As more innovation and implementations spin up in the always dynamic worlds of #IoT and the Industrial IoT ( #IIoT ), the inevitable evolution to more real time, or at least near real time applications becomes clearer each day. For real time applications to work, events processing requires a fine tuning of every element in the architecture, from the devices at the edge to gateways and networks, clouds and applications, all which must be programmed and secured, and made economically feasible, relative to the value "real time" delivers. Real time data networks have been around and running mission critical, connected systems for many decades, most prevalently in financial services (exchanges), transportation (traffic control), public safety (command and control), eCommerce (payments and recommendation engines), telecom (interconnection, operating and bill systems, network management) and are part of the "nervous system" of the primarily digital world. As more and more physical things become digitized, entrepreneurs are taking up the challenge of making data from those machines and transactions more valuable through automation towards "zero touch" provisioning, other actions once devices are securely provisioned, and ongoing management of buildings, factories, thoroughfares, and more. The benefits are easy to imagine, but putting real time systems in place given the complexity and variety of IoT and IIoT solutions is more challenging than product developers and their engineers might imagine. Real time "event driven" stream-based processing is a vast topic, and while we have ample precedent and understanding of what elements must be thought through (data import, processing, security, storage, extraction and integration with analytics engines to feed applications, for example), experts are exercising caution even as they, too, see the potential of massive, intelligent automation. Three such experts - Don DeLoach, Emil Berthelsen and Wael Elrifai - have joined together to publish a book - The Future of IoT - about to be released online and hit the physical bookshelves that are left - and one of the many important topics they grapple with is event-driven, publish and subscribe architectures. DeLoach, one of the world's experts in real time events processing who has built and sold a number of companies whose products have supported many of the world's largest banks, scientific organizations, payments processing platforms and more, put it this way: As IoT enabled products evolved, one of the key and notable characteristics of most of these products were the alerting and triggering mechanisms. One would correctly view this as an event driven product where the “event”, be it a temperature drop or a door state condition, or any other measurable piece of information, is interpreted by a rules engine or workflow engine to “respond” by issuing an alarm or some type of notification, or by triggering some other action, like shutting off the power to itself or switching speeds to “respond” to a detected condition. In these IoT enabled products, the “event” is the absorption of the message coming from a given sensor reflecting the state of a given product. When the state hits a certain threshold, the prescribed action is engaged. This describes the typical IoT enabled product. As IoT emerged and until recently, many might describe this is “event driven IoT”, but really it means “Event driven IoT products”. However, DeLoach, who recently sold the company he was running (Infobright), was inspired to write the book, along with Emil Berthlesen, a Research Director at Gartner, and Wael Elrifai, a technologist and strategist for the Pentaho division of Hitachi, to focus in on the IoT-enabled enterprise, vs. IoT products per se. In the book, the subtitle of which is "Leveraging the shift to a data centric world," the authors describe the evolution of the IoT market as driven by the shift in focus from a single IoT enabled product to large enterprises operating with hundreds or thousands of IoT enabled products, resulting in tens or even hundreds of thousands of endpoints under management, and how the enterprise can achieve far greater leverage by looking at IoT product generated data in correlation with the data from the other IoT enabled products. "IoT and IIoT will no longer be freestanding systems, islands outside the growing data lakes rising up in enterprises," DeLoach said. "The top CIOs, IT teams, developers and DevOps groups are getting better and better at understanding data and planning data architectures which can now include connected things. This dramatically enriched set of information can be used to drive far greater insight, taking the form visibility into the current state of the operation - what is going on - into the cause and effect - why it is going on - and into the probability of certain outcomes to happen in the future - what will be happening." Most notably, DeLoach is seeing a tremendous amount of demand for adaptability, and the use of real time data to respond to changing conditions. "Machine Learning is becoming a key part of the toolset used by many enterprises, especially for the Internet of Things,' DeLoach said. "But the height of success for IoT is not adapting to conditions within a silo (or product), but rather adapting to the holistic conditions to optimize on an enterprise level. This requires leveraging the data across the enterprise, taking into account the range of IoT enabled assets, the enterprise systems (MRP, POS, CRM, etc), and perhaps even external data from supply chain partners, city or other government supplied data (weather, traffic, etc.) or other data used to enrich the datasets used to gain insight." DeLoach makes a clear distinction between devices, or end-points that can deliver real time information, and event driven IoT systems, which make a big dent in what it costs to run significant aspects of a business, systems that have very clear "ROIs" necessary to fund not just "things" but entire data architectures supported by the necessary infrastructure, including secure, resilient networks, to ensure what becomes automated, including with Artificial Intelligence processing built in, doesn't go haywire and cannot be compromised by cyber criminals. The theme of the "event driven IoT enterprise" runs throughout the book, which focuses a great deal on how the connection and management of systems on the context of the treatment and propagation of the data will impact the digitally transformed enterprises of the future. When asked what drove him to contribute to the co-authored volume, another IoT expert, Wael Elrifai said," We were inspired to write this book now after working with many enterprise companies on large-scale IoT projects. The nature of IoT is that projects tend to be very long term - much more so than an IT leader's average tenure in a company - and complex. Without knowing IoT's direction of travel, IT strategists risk designing architecture and governance models that don't fully leverage data's value or in worst case, must be re-engineered completely. Our hope is to help resolve that challenge for early adopters." The book, which ventures well beyond real time event stream processing, is interesting in that it combines the way of thinking of a CEO, operating companies bringing products to the IoT and IIoT world, a business strategist whose role is to understand and guide complex, long-term projects, and an industry analyst who meets with dozens of developers across the spectrum, to dig deeply into their particular offerings, be they sensors, equipment, infrastructure, networks, data, analytics, cloud and more and more AI and deep machine learning. Emil Berthelsen has published for years on trends in IoT, and now as part of Gartner's global force, has an even bigger playing field to work in. “The focus in M2M and IoT has for a long time been the technology, and rightly so,” Berthelsen said. “Enterprises and technology service providers continue to face challenges in this fragmented and heterogeneous environment. What has become a new challenge for companies is the transformation from highly product-centric businesses to service and data-driven organizations. This means that the utility value of data and how it is managed must become a critical priority. Companies will need to implement changes in both internal and external structures, business models and resources to meet the future of IoT” Cloud publish and subscribe scenarios, as more and more messaging makes its way into the real time IoT and IIoT, is another topic covered in the book, and one which we will be sharing in a second article in the near future. "Cloud publish and subscribe messaging systems are on the rise, and must take into consideration the requirements for incoming data streams as well as changes to how applications will be automated, including on a round trip," DeLoach said. "When we start to get into the world of distance automation,” DeLoach continued, “we need to take a close look at what can be processed at the edge, in a 'first receiver' model, and what needs to move to the cloud for more distributed applications, and more robust analysis not necessarily needed to make sure the temperature and humidity levels are optimal, for example. When sensors respond to events in the practical, physical world, enterprise systems need to be prepared for swarms and floods, streaming into data ingestion, storage and manipulation systems." Data approximation techniques will be critical, DeLoach says, to make all the potential realistically possible. "As processing, security, transport, licensing and other economic factors weigh in, progress versus perfection may win the day." It's not easy for engineers to accept gray areas, but when we apply the right gray matter to solving for the balance of utility and cost, proximity and approximation become calculus, not math.
http://www.iotevolutionworld.com/iot/articles/433237-future-iot-moving-iot-iiot-into-real-time.htm
No comments:
Post a Comment