OptoBlog

The rise of the IoT middleware platform

Posted by Matt Newton on Feb 21, 2017 8:00:00 AM

IIoT Applications Require New Middleware That Connects OT and ITIn a recent developer survey conducted by the Eclipse IoT Working Group, IEEE IoT, and Agile IoT, 40.8% of respondents indicated the Internet of Things (IoT) solutions they were either currently building or planning to build were IoT platforms or IoT middleware.

Almost every large IT or OT company today has some form of IoT platform or middleware, with the current number of self-described IoT platforms exceeding 150.

So what are IoT platforms and middleware? They're the software that must exist between physical devices (sensors, actuators, relays, etc.) or data endpoints, and higher-level software applications like artificial intelligence, predictive analytics, and cognitive computing.

IIoT platforms and middleware move data between the physical and digital realms and provide software resources powerful enough to cope with the big data generated from the billions of IIoT devices the industry is planning to ship.

Get IoT Resources

While middleware is not a new technology, the IIoT has given rise to a need for a new type of middleware specifically designed for IIoT applications. That's because IIoT applications require a middleware platform that addresses new concerns, including:

  • Scalability
  • Edge computing
  • Efficient communication architectures
  • Protocol support
  • Cognitive computing
Let's take a look at these new concerns and why they're important to the IIoT.

IIoT middleware platforms must communicate thousands of devices

Scalability 

The sheer volume of devices and other data endpoints involved in IIoT applications requires advanced and flexible scalability.

IIoT platforms and middleware need to be optimized to support tens of thousands of devices, all trying to exchange not just information with a central hub server or application but with each of the other devices in the application.

Addressing, configuring, and managing thousands—even tens of thousandsof devices must be accounted for.

Edge computing 

Centralized intelligence and control topologies are being reevaluated in favor of distributed architectures, with intelligence pushed into each edge device or data endpoint.

IIoT applications involve thousands of devices intercommunicating, often with the requirement for near-real-time communication and control between devices. That's a lot of data!

Edge computing uses intelligence at the edge of the network to decrease network latency, deliver real-time control and monitoring, and offer report by exception to reduce data volume.
Get The Edge Computing Primer

Efficient communication architectures 

The traditional client/server architecture of the Internet is not scalable in IIoT applications. Instead, a brokered publish/subscribe architecture is more bandwidth-efficient, as data is transmitted only when a change occurs.

Instead of polling a pump, relay, or other end device for its status every second, the end device reports its status only when a change occurs. This architectural shift is required to reduce network latency and improve real-time communication speed between devices and data endpoints.

One publish/subscribe protocol that has a lot of buzz around it right now is MQTT.
mqttorg_666557.png

Protocol support

Combining OT and IT technologies requires wide support for very different OT and IT protocols. At least in the short run, both OT and IT protocols need to be translated through middleware, so that OT devices can communicate with IT devices and software.

In the long run, it is likely that OT devices will adopt IT protocols and communication standards, much as they’ve already adopted Ethernet and TCP/IP as the main bus and data transmission protocols. But there is still a massive installed base of legacy industrial systems that will always require some form of middleware for protocol conversion and connectivity to the IIoT.anatomy-1751201-2.png

Cognitive computing

The key value-add of IIoT applications is predictive analytics.

Knowing when a part is going to fail before it actually fails can bring almost immeasurable value to IIoT applications through reduced truck rolls, safety improvements, and optimizing overall equipment efficiency (OEE).

The basis of predictive analytics is cognitive computing—essentially, computers that mimic the way the human brain works. For IIoT platforms to perform predictive analytics, they’ll need support for cognitive computing.

Where we're headed

Right now—and for some years to come—our root problem is that IIoT applications inherently require connecting legacy systems and devices to cutting-edge IT systems. And a massive gap exists in technology, communication protocols, and standards between equipment designed several decades ago and the equipment shipping today. That's the gap IoT middleware is trying to fill.

But IIoT applications are not and should not be designed as a one-size-fits-all solution. Choosing the right IIoT platform depends on what you are trying to achieve in your application.

Sign up for the Opto 22 blog to get our survey of the IIoT platform landscape for 2017.

Subscribe to the OptoBlog

Topics: groov, Remote monitoring, Tips, IoT, Networking, IIoT, Node-RED, Data acquisition

Written by Matt Newton

    Subscribe to Email Updates

    Recent Posts

    Posts by Topic

    see all