Software-Defined Networks Tame IoT Complexity
High-speed, reliable, and flexible networks provide the foundation for transporting information from connected devices to backend business systems. The diversity of IoT applications requires organizations to consider a wide-range of networking options, including traditional Ethernet networks, high-speed WiFi infrastructures, Bluetooth connections, cellular services, and emerging wireless standards like Sigfox and OnRamp.
To tame the management complexity threatened by large and diverse IoT networks, growing numbers of IT administrators may look to software-defined networking (SDN) and network functions virtualization (NFV), which can centrally control communications pipelines. IDC cites IoT as one of the key drivers that may fuel the worldwide SDN market's rise to $8 billion by 2018.
(Image: Open Networking Foundation)
Processors Expand IoT Options
IoT architects have a variety of processor options for powering sensors, RFID devices, and other IoT endpoints. Microcontroller units combine a processor, memory, and an input/output controller within a single chip. Some vendors, such as ARM, offer IoT platforms that include a microcontroller, an operating system, a device server, and related tools in an integrated package.
Silicon on chip (SoC) is another option for chip-level processing, but typically with more powerful processors and larger amounts of memory. Intel's Quark SoC X1000 series uses an instruction set that's compatible with 32-bit Intel Pentium processors and is part of Intel's platform of IoT hardware and software.
Enterprises can expect to see more solutions that bundle processors with related capabilities and services as semiconductor companies look beyond just silicon to fully capitalize on IoT, according to McKinsey & Co.. Ancillary capabilities that are generating the most interest include software, security, and systems integration, the management consulting firm says.
Storage For All That Data
Once endpoints gather information and send it to backend systems, organizations have two fundamental choices for storing all the data: traditional on-premises storehouses or cloud-based services. Traditional storage arrays come with a healthy upfront capital investment and ongoing hardware purchases over time as data volumes grow. They also require dedicated staff to manage the resources.
But some industry analysts believe the long-term costs for managing what may be petabytes of data may be lower with on-site hardware than with cloud, especially for services that charge fees for transferring information out of the cloud. By contrast, cloud services can reduce capital investment requirements and relieve storage management requirements for internal IT staffs.
Gateways Connect Devices
IoT gateways enable legacy equipment that's been upgraded with sensors or other data-collection devices to communicate with backend management systems. For example, with the help of gateways, machinery on a production line can relay information that alerts service technicians to impending maintenance problems.
IoT gateways typically fall into two categories: simple units that communicate raw streams of data or intelligent gateways that perform some initial data analysis so they can transmit only a subset of the full data stream. The result: the pricier intelligent gateways reduce the volume of data flowing across the network to management systems and help organizations identify high-value information.
(Image: IBM Point of View, April 2015)
The volume and diversity of IoT data requires a mix of databases to manage information and serve it up to analysts in the right format. Familiar SQL databases handle data stored in conventional rows and columns for easy access and analysis. NoSQL databases can accept incoming data at higher rates than their more traditional cousins -- 50,000 vs. 5,000 inserts per second, according to some estimates. Unlike SQL, NoSQL technology supports both structured and unstructured data. NoSQL databases may be paired with Hadoop data processing systems, which perform extract, transform, and load activities.
Hybrid databases combine in-memory and disk-based storage to balance high performance and budgetary considerations. "In some cases, SQL and hybrid databases will meet the requirements of businesses, and in other cases, NoSQL databases will be the way forward, a report by Machina Research concludes.
(Image: Machina Research)
Analytics In A Range Of Options
With advanced analytics, users of IoT solutions can slice and dice high volumes of data to optimize the systems being monitored, predict trends, and perform simulations. IT administrators have a few options for implementing sophisticated analytics. Gartner breaks down the choices into three main categories.
The first is to install an advanced analytics platform that provides a suite of tools for creating analytic models. Leaders in this category, according to Gartner, include SAS, IBM, KNIME, and RapidMiner. Gartner notes that increasing demand for analytics has put highly trained data scientists in short supply. The result is that analytics platforms offer higher levels of automation and self-service tools for use by what the research firm calls "citizen data scientists."
Alternatives to installing an analytics platform include cloud-based services.
Security Tailored To Architecture
A recent study conducted by IDC and sponsored by Cisco found that most decision makers recognize that IoT security is inherently different than traditional IT security. However, some firms do not understand the high-level protocols and technologies required to secure IoT solutions, which prevents them from taking the necessary measures toward IoT security.
In "IBM Point of View: Internet of Things (IoT) Security," the company points out that security strategies will differ depending on each IoT architecture. For example, IT administrators must embed security within each device that directly connects to the Internet. In other cases, devices may connect to a gateway, which would then take on the primary security responsibility.
The importance of security helps explain the spate of product introductions and acquisitions in the market. For example, silicon vendor ARM recently acquired Israel-based Sansa Security, a provider of hardware security IP and software for IoT components. The technology adds to ARM's existing security portfolio.
(Image: IBM Point of View, April 2015)