Follow Datanami:
April 3, 2018

Built to Last: Laying a Framework for IoT with Enterprise Architecture

André Christ

(BeeBright/Shutterstock)

IoT is massive — both in opportunity for business and volume of data created. Over 90% of the data in the world has been created in the past two years, and the current output of data is roughly 2.5 quintillion bytes a day. As more organizations move from initial experimentation phase to full deployment of IoT applications, the data deluge will continue to plague organizations as they try to capture, process, and act on the immense volume of information.

IT organizations steeped in legacy infrastructure are ill-prepared. Rigid architectures and traditional data models are quickly becoming obsolete, as they weren’t built to handle the speed and agility of new programs like machine learning algorithms that are critical to deriving real value from IoT.

As with all nascent systems and processes, the path to an IoT-friendly infrastructure is riddled with obstacles. Chief among them are integration, complexity, security, and transparency. Let’s discuss each of these in turn.

Obstacle: Integration 

Solution: Ensure interoperability of applications

 Gartner research analysts claim that through 2018, half the cost of implementing IoT solutions will be spent integrating various IoT components with each other and back-end systems. A big challenge organizations will face is enterprise architecture (EA) system integration to support the volume of data. Incoming data will overwhelm existing systems causing errors, failures, and downtime, which is why data of heterogeneous connections, devices, and sensors needs to be integrated.

The data flowing from devices to IoT cloud platforms to analytics systems needs to be transparent. In order to ensure transparency, data silos need to be avoided so the right teams can better process and analyze data. Through interoperability, data integration enables organizations to give value to the relevant data, which can then be tapped for further analysis and service design.

Obstacle: Complexity

Solution: Drive and track transformation progress in projects

IoT transformation projects are complex and require careful planning and tracking against progress. Having an IoT roadmap will keep you from adding valueless technology to your landscape. Enterprise architects should be in the driver’s seat, and lead when identifying conflicts in requirements between different projects regarding the same applications.

Planning and tracking the transformation process can cut down the time of the entire process of successfully deploying the IoT-supported system. Here, enterprise architects can easily track the phase-ins of new applications and retirements of legacy applications, and plan for scenarios of the application landscape to future-proof the organizations’ system.

Obstacle: Security

Solution: Deep visibility and ability to sense/respond quickly

With data breaches occurring almost weekly, security is a crucial issue and proves to be a significant challenge for IoT. One of the biggest and most impactful costs of integrating an EA system that supports IoT is the potential security risks to the organization if left exposed. Having a high number of new endpoints requires a high focus on security.

EA’s undergoing IoT integration need to evaluate the technology risk for applications and business capabilities based on the organization’s underlying IT components. Identifying IT components to be replaced can be a great step in filling in any open gaps and mitigating security risks.

Obstacle: Disparate stakeholders

Solution: Access and transparency for all relevant business units

Establishing and extracting relevant data for enterprise architecture and future business decisions can help the entire organization better leverage the power of IoT data. By identifying relevant KPIs to track the status of the IoT infrastructure, EA’s can get a better understanding of how incoming data can be optimized to support every stakeholder’s goals. Applying IoT real-time and time-based metrics as a single-point of information can drive impactful decision-support faster than legacy architecture systems of the past.

Alignment is mission-critical when undertaking an architecture transformation. A business capability map will give the necessary visibility and foster the tight collaboration between IT and business needed for successful deployment. Capability maps help evaluate and manage the entire application landscape providing a bird’s-eye view into which existing applications can and can’t support IoT.

Based on the defined criteria laid out in the business capabilities map, teams can evaluate the quality of support of its existing applications can handle, develop requirements for these applications, retire legacy applications, and identify the need for new, enhanced applications before IoT support systems begin. Taking the time to identify and evaluate your application landscape can ensure any gaps in support are taken care of ahead of deployment, as well as pinpoint any areas to improve overall efficiency in across the application landscape and entire organization.

Now that we’ve covered our bases, and have shown why and how IoT can affect organizations, we need to explain how EA’s can successfully move from the rigid legacy infrastructures of the past, and deploy a system that supports your organization’s bottom line now and in the future.

The Four-Stage Architecture of an IoT System

Stage 1: Making sense of the sensors

The connection between the the digital and physical world is realized through wireless sensors and actuators. Objects equipped with sensors, actuators, processors, and transceivers  communicate in a network — the Internet of Things. The architecture of an IoT system consists of several layers — starting with physical devices.

(mailsonpignata/Shutterstock)

Wireless sensors and actuators collect incredible amounts of user data — from sign-on times, level and hours of usage, location statistics and more. This raw, inbound data can easily overwhelm and clutter core IT infrastructures. Sensors and actuators work together to produce data. For example, consider a cell phone – the camera and microphone are sensors, whereas the speaker and screen are the actuators. Actuators also have the ability to produce physical changes based on data from a sensor, such as shutting off a device, moving equipment, and more.

Industry estimates predict billions of IoT-based sensors will surround us by 2020, with the average household creating enough data to fill more than 300 32gb iPhones. The weight of unneeded information flow will tax the architecture to a breaking point and make analysis less efficient. Which is why the first step to building an IoT-centered architecture is understanding what data is and is not needed, and planning accordingly.

Stage 2: Opening the Internet Gateway

Data collected from sensors is raw. To be useful, it must be aggregated and converted into digital streams for advanced processing. This is the second layer of an IoT architecture system. 

Using a data acquisition system (DAS or DAQ), information from sensors and actuators becomes digital — converting analog waveforms into digital values for processing and connection. Essentially, this is the process of sampling signals that measure real world physical conditions and convert the resulting samples into digital numeric values that can be manipulated by a computer.

The DAS connects to the sensor network, aggregates outputs, and performs the analog-to-digital conversion. The Internet gateway receives aggregated and digitized data and routes it over to Wifi, wired LANs, the Internet, or to edge systems for deeper processing.

Stage 3: Computing at the Edge

In nearly every IoT use case, edge computing is deployed to streamline the flow of traffic from devices and provide real-time local data analysis. In moving data away from centralized points to the periphery of the network, you significantly decrease the volume of data that must be moved and the distance that data must travel. This, in turn, reduces transmission costs, shrinks latency, and improves QoS. By effectively eliminating the core computing environment, you also avoid a potential point of failure.

In the event that data requires further processing before it enters the data center or the cloud, you’ll need an edge IT system. The preprocessing data systems works at the edge of the network where IoT connects the physical world and the cloud. Edge IT systems use the processing power of IoT devices to filter, pre-process, aggregate, or score IoT data. These applications also use the power and flexibility of Cloud services to run complex analytics on data and, in a feedback loop, support decisions and actions in the physical world.

This stage is a fundamental part of the process and represents the strong and seamless integration between IoT and cloud — between the physical world and the world of computation. Edge systems also lessen the brunt of labor on organizations’ core IT infrastructures. IoT data can easily overspend network bandwidth, overwhelming data center resources and increase the risk of security concerns, stressed networks, and computations lag. Pre-processing can easily mitigate the stresses and risks to your infrastructure, while gathering meaningful and actionable data for your organization.

Consider using machine learning techniques at the edge to scan for anomalies and impending maintenance issues that will require immediate attention. From there, you can trigger visualization technology to present information in easily digestible maps and graphs. However, it’s important to keep in mind that big data analytics and machine learning processes require a lot of power. Establishing a separate cloud-based database for your data science and machine learning tasks, so that queries and calculations do not impact the product or service day-to-day processes or stress your core IT infrastructure.

Stage 4: Managing, securing and storing 

Once data from the edge is delivered to the core environment (data center or cloud), you can then do proper analysis and management and set secure storage. While this final round of processing does take a bit longer to deliver results, the delivery will include deeper analysis and actionable results. After data has been aggregated, cleaned, and surveyed, the information can be fed to the server to be analyzed and applied to new products and services.

The winners in IoT will build a firm, agile, and scalable enterprise architecture (EA) that supports the demands of IoT now and in the future. Leveraging the power of IoT can produce information and insights in a manner new to almost every organization. From enhancing digital workflows and increasing collaboration and efficiency, to unveiling new insights unknown to the organization before. With the proper architecture in place, EA’s can extract significant value from big data, helping organizations make impactful data-driven decisions.

About the author: André Christ worked for years as a management consultant at a leading consulting company before he founded LeanIX. As a consultant, André advised CIOs and IT leaders on strategic topics, such as IT complexity reduction or defining service and cost models for global data centers. During his studies of Information Science in Muenster and Montpellier, he worked as a freelancer creating modern software architectures for both startups and large enterprises.

Related Items:

Learning IoT Lessons the Big Data Way

Data Opportunities Slipping Through the Cracks

Finding Hidden Value in IoT Analytics

Datanami