A core component of any successful large or small IoT implementation is an effective device management solution. An IoT device management platform (DMP) is an offering that provides device lifecycle management functionality associated with the deployment and management of IoT assets. Typical IoT assets include IoT gateways, retrofitted and new industrial equipment and low-powered Linux-based devices. Typical DMP functionality includes asset provisioning, firmware upgrades, security patching, alerting and reporting on specific metrics associated with IoT assets. In short, a DMP provides the glue that ties together the assets and devices comprising the physical layer of an IoT solution.
While many vendors offer effective DMPs, there are 3 weaknesses common to nearly all DMPs: lack of edge management functionality, lack of device agent productization and lack of external integration.
Weakness #1: DMPs often do not have strong edge management functionality. Without a strong edge management offering within a DMP, customers become over reliant on external cloud or network services to drive local decision making. Edge management is the process by which a DMP exercises control over a connected and provisioned edge device such as an IoT gateway or other IoT asset providing on-premises compute capabilities. An effective edge management implementation should enable remote configuration, remote software management, connectivity management, edge deployment and edge autonomy. Typical IoT deployments are heavily reliant upon cloud-services. An effective edge management solution provided by a DMP lets customers visualize, process and take action on ingested data even when cloud connectivity is absent or intermittent. For example, in manufacturing or environments where IoT devices interact with physical machines, the added latency and reliance on external connectivity can significantly impede efficient operations, and moreover, edge devices without strong autonomy can become disabled during periods of network outage. Having strong edge management functionality within a DMP corrects this weakness. For more details on the specific requirements for effective edge management as well as current leading vendors, please refer to our 2018 IoT Edge Scorecard.
Weakness #2: The second common DMP weakness is the lack of productized device agents. Providing a productized device agent allows customers to quickly test the full functionality of an DMP without investing tens or hundreds of developer hours just to trial an IoT device. A device agent is a piece of software code, typically deployed as a containerized service or a binary large object (BLOB), that runs on an IoT device or gateway and enables bidirectional communication with the DMP. A device agent enables many of the typical IoT lifecycle management capabilities such as firmware or software updates, device-to-cloud protocol support and cloud-to-device command support. A device agent may also support normalization or translation of ingested data or configurable northbound forwarding of data streams. Effectively, these device agents are the common layer of compatibility that allows an IoT device to integrate with the northbound DMP solution.
While many vendors offer SDKs or other toolsets to enable customers to compile and deploy their own device agents, much less frequently do they offer pre-compiled productized device agents for commonly used hardware or device applications. Given that device integration is one of the most fundamental requirements of an IoT deployment, this means that one of the first implementation hurdles for customers is to configure, compile and design a compatible device agent for devices requiring platform integration. Further, while many platforms purport to offer over-the-air software or firmware updates or other typical DMP functionalities, very few provide this features out-of-the-box within their sample device agents, if any sample device agent is provided at all. This leads to significant additional complexity, as customers must invest developer effort not only to deploy the core device agent, but also manually implement file management, software updates and other features offered by the DMP. MachNation believes that DMP vendors should offer a selection of fully-featured, ready-to-deploy device agents for common hardwares in addition to a generic, easily-implemented device agent SDK.
Weakness #3: The third common DMP weakness is the lack of options typically provided to enable cloud-to-cloud or machine-to-machine integration with external applications or services. An effective DMP must offer multiple methods to expose ingested IoT data streams to external systems, thereby simplifying customer implementation of enterprise application integration and enabling faster and less costly solution deployment. Most DMP implementations enable ingested data to flow from the DMP into enterprise applications such as CRM or ERP systems. This allows customers to generate billing metrics, monitor long-term KPIs, interface with inventory management systems and connect client orders in ERP or CRM systems with on-premises equipment. Ingested IoT data is of little value until it is given an appropriate business context:most IoT solutions will leverage existing enterprise systems to give IoT data this business context. Therefore, it is of critical importance that a vendor’s DMP provide abundant methods for integration with external systems, preferably in a fully productized manner that removes the need for custom middleware between the DMP and a given enterprise application. While many vendors expose RESTful or MQTT endpoints that enable polling of ingested data, fewer offer the ability to stream real-time DMP data through a push mechanism. Even fewer platforms offer the ability for productized integration with CRM or ERP systems, thereby requiring expensive or complex middleware to handle the integration of enterprise applications and the DMP.
A DMP is a key aspect to IoT solution success. While some customers may attempt to build their own implementations, this is often a costly and complicating mistake. Instead, choosing a fully featured DMP offering that provides these key functionalities out-of-the-box saves time and development effort