Agencies Need ‘Holistic’ Data Architecture for Internet of Things

Internet of Things IoT Data Architecture diagram

Since the Internet of Things (IoT) is a living, growing, evolving web of networks, organizations need a data architecture that can constantly adapt to its dynamic nature. For government agencies, that also means building a “holistic” data architecture that can accommodate both real-time streaming data and data stored in traditional enterprise databases, industry experts say.

There is “a whole plethora of publishers that are creating data that needs to be stored, processed, and computed to provide information and intelligence” to agency managers, from smart meters for energy management to sensors for weather forecasting, Shaun Bierweiler, vice president of U.S. Public Sector with Hortonworks, said during an interview with MeriTalk.

“We see a lot of that in the government space, and I would argue that they [government agencies] are leading the way in that realm” of tapping into IoT and different data sources.

For instance, the General Services Administration has deployed smart sensors within GSA buildings to monitor and improve energy efficiency. The Census Bureau is now deploying handheld devices and other IoT tools to turn the gathering of census information–traditionally a manual and paper-driven process–into a more real-time, digital process, he noted.

Many Federal agencies want to create data and ingest that into data architectures, potentially marrying IoT data with enterprise data–structured and unstructured–to provide desirable analysis and outcomes, Bierweiler said.

The Internet of Things is not just a network, rather it is a web of networks with different hierarchical levels. Information within these levels can be streamed in either direction–from the device to a computational environment where it can be filtered and analyzed–and then back to the device by analytical models. The models can send back information, such as instructing a sensor to lower the temperature on a device, according to David Loshin, president of Knowledge Integrity, Inc.

Loshin described some of the characteristics of an IoT architecture during a webinar, “The Fundamentals of IoT Data Architecture,” on Aug. 29, hosted by TDWI and sponsored by Hortonworks. Hortonworks Connected Data Platform allows enterprises to seamlessly collect and analyze streaming data from the IoT devices and perform real-time, large-scale, and high-speed analytics on large volumes of data to generate immediate insights.

Organizations should build a data architecture that supports an IoT environment’s ability to absorb and make use of data streams in the right context to achieve the desired outcomes, Loshin said. As a result, certain steps should be considered when building an IoT architecture, such as:

  1. Build a data architecture that adapts to the dynamic nature of the network topology. New devices are constantly being added into the network and information is subject to change without notice.
  2. Simplify the integration of IoT devices. “As new data streams come in, you don’t want to wait a long time to figure out what data streams mean and want is being generated” by the devices, Loshin said.
  3. Facilitate agile development. “If we want to be reactive in an informed way to changes in the network, we want to make use of what we’ve learned and to rapidly integrate those changes or modifications into our holistic system for doing ingesting of analytical models,” Loshin noted.
  4. Ingest and integrate streaming data in a way that is thoughtful and knowledgeable about the context of how the data originates.
  5. Enable analytics, not just analysis of what data is coming, but the ability to take outcomes of analytical models and deploy them to the right place in the network.
  6. Embed governance and levels of control in the network to monitor what is going on and identify when you are getting the level of improvement you expect, beyond what you expect, or if it is missing expectations.

As agencies consider an IoT architecture or any data architecture/data platform, information and network security should be top of mind, Bierweiler said. “Having security baked-in as a fore-thought, not as an afterthought, is very important.”

Recent