As many companies are continuing to implement “smart”, network-aware devices in industries such as manufacturing, oil & gas, energy & utilities, transportation & logistics, etc. the desire for remote monitoring and management of these devices has grown. There is immense business value in delivering the enhanced control, awareness, and automation that such Internet of Things (IoT) solutions can offer.
In a previous article, we discussed a typical industrial IoT architecture (see Figure 1) and how the MQTT standard can enable communications between edge sites and a centralized cloud-based solution to enable the most value from an IoT solution.
Figure 1 – Typical Industrial IoT Architecture
In this article, we’ll take that architecture and enhance it to include the capability to process data on the edge. This enhanced architecture, while not required for many implementations, offers several benefits to the overall solution:
- It allows for real-time telemetry analysis and automated decisioning to improve the efficiency of the devices running at each edge location. Similar processing can take place in the cloud, but increased latency to perform such processing in the centralized environment can reduce your ability to react quickly to mission-critical events.
- Data can be enriched and aggregated at the edge to maximize the data transfer efficiency to the cloud. This is especially critical for edge sites that have intermittent connectivity and cost-prohibitive connectivity costs.
- Cloud infrastructure costs can be reduced. By keeping some amount of process at the edge site on owned hardware, you can potentially reduce the data transfer, storage, and compute costs in your cloud environment(s).
These benefits can have a profound impact on the overall capabilities and operational effectiveness of your IoT solution. In designing the solution, it will be important to see if the benefits gained outweigh the added scope and complexity on the edge implementation. Let’s take a look at how enabling such edge processing would impact our design. Figure 2 shows a high-level view of how such an architecture would look.
Figure 2 – Industrial IoT Architecture w/ Edge Processing
As you can see, the cloud portion of the architecture remains similar to the original architecture. What’s changed is that we’ve added two additional components on the edge:
- Edge Broker: This component is responsible for several aspects of the solution. First, it manages the connectivity with the Controller so that device telemetry data can be communicated (typically using MQTT). Second, it provides a way for the Edge Data Processing engine to subscribe to the data streams coming from the Controller and publish data and commands back onto the message bus. Third, it manages the connectivity to the cloud Gateway so that data can be streamed into that environment for processing and access.
- Edge Data Processing: This component allows for the edge site to validate enrich, and aggregate the data coming from the devices. Typically it will have one or more data stores available to it for tracking events and trends in the telemetry data coming through the Edge Broker. This allows for real-time decisions to be made and commands or alerts to be generated with minimal delay.
In some instances, the edge architecture may also include an administration console running at each site that allows technicians to view the status of the edge components. However, since many Controller devices also provide similar capabilities, it’s not always a necessary part of the IoT solution.
If you’re consider designing an IoT solution and are unsure of whether having data processing on the edge is right for you, Anexinet can help. We’ve architected and implemented these solutions for multiple clients and can help you design the best solution for you and your company.