Handle industrial IoT data end-to-end
One of the first markets for the Internet of Things (IoT) that analysts expect to gain real traction is the industrial sector. The opportunities for predictive maintenance and increased production efficiency that instrumenting production equipment will provide have great promise for offering significant ROI on IoT deployment. But to be most effective, these industrial IoT systems will need to integrate with existing factory and management IT networks, which have completely different needs. It appears, however, that new DDS (data distribution service) technology may be able to bridge these networks with a common protocol.
Factory networks typically produce and use continuous streaming data as part of their operational technology focus. The data comes at the speed of physical processes, gets routed to many points in the system, and requires a real-time response. The compute resources available in these networks are relatively fixed, without an ability to quickly bring new resources to bear.
The management IT networks, on the other hand, are designed for transaction processing. Data comes in batches and processing is periodic. Processing only needs to be fast enough to meet the needs of human operators, and processing resources within the network can typically be reassigned or added to from the cloud as needed for load balancing.
Existing networks in industrial settings use differing connectivity products and standards. At the edge the factory equipment might use DDS or a proprietary, industry-specific protocol. Linking from the edge to the cloud, DDS, MQTT, HTTP, or SMPP might be involved. Within the management IT network, messaging protocols like ESB (enterprise service bus), AMQP (advanced message queuing protocol) or other message-oriented middleware (MOM) are at work. These protocols typically move data from point to point within the network through a queue.
There hasn't been one single protocol that can handle data end-to-end, from the edge to the IT system. Instead, bridging these disparate network needs has typically required the use of gateways and such to translate protocols and to package and move data. As the solutions tend to be application-specific, with different APIs, it is difficult to share or move applications across an industry.
But that may be changing. Real-Time Innovations (RTI) has recently released new version 5.2 of its Connext DDS software that it claims will allow edge-to-cloud connectivity, providing a single communications mechanism that will meet both industrial and IT network needs. The publish-subscribe method of DDS already handles the edge needs and can provide the link to the cloud. What has been added in version 5.2 is the ability to support the point-to-point queuing needed in the cloud interactions. This addition allows developers to use a single technology end-to-end, with consistent security model and API at all layers, while meeting the information-handling needs of every stage.
A useful addition to the toolkit supporting Connext DDS is the ability to observe the data flowing through the system. The toolkit already had the ability to provide system-level visualization of the dataflow paths. This addition also allows developers to see the actual data being moved, for debug and control purposes.
Not all existing industrial systems use the publish/subscribe model of data flow that DDS supports, of course. But the industry is increasingly moving that way as the model greatly simplifies the expansion of networks when adding new data sources and consumers. If the RTI extensions to its DDS pay off as promised, then industrial applications of the IoT are poised to take off dramatically.