A new breed of Analytic Gateways are arming IoT solution providers with more intelligence, enhanced connectivity, reliability, robustness and security to bring the right analytics to where it’s needed most, at the edge.With potentially hundreds of petabytes of data coming from billions of connected devices by 2020, architects of the Internet of Things (IoT) are grappling with how to process the data flood to avoid latencies, ensure security and optimize revenue-generating potential.
A centralized approach, with all the data streaming to the cloud for processing and analysis, sounds neat and manageable, but there are far too many dependencies: robust, reliable connectivity, end-to-end security, and of course server availability. A break along the way can delay action at the edge, which isn’t good if a motor needs to be shut down or a customer’s profile isn’t sent in time to a retail store to maximize engagement.
The alternative is a more decentralized approach, where actions and decisions happen at the edge (factory floor or point of sale), using structured data stored locally in a database, combined with incoming real-time unstructured data. This requires a good degree of processing in the edge device, and a balancing act between what to do locally versus what’s best to be done in the cloud.
However, this decentralized approach does enable faster reaction times (lower latencies) and directly enhances security by decreasing the amount of data being transferred back and forth. That said, there is a cost factor associated with adding more intelligence and analytics at a remote IoT gateway, which now goes from being an important enabling dot to being a really big, critical dot. There also needs to be more emphasis on security within the gateway, as it’s no longer an aggregation and distribution point, but now stores more information, and the fact that it is more intelligent means a breach can have a greater impact on the final outcome than ever before.
It’s with all this in mind, and a bit more to boot, that companies like Dell have started a trend toward what it calls IoT Analytic Gateways that have smarts built in to handle a good chunk of the data analysis, as well as much more robust connectivity for multiple protocols from many different devices.
Figure 1. The Dell Edge Gateway 5000 Series is a lead-edge example of how to balance power, performance, flexibility and connectivity to enable effective edge analytics. (Source: Dell)
For example, the Dell Edge Gateway 5000 Series, based on either the two-core Intel® Atom™ processor E3825 running at 1.45 GHz or the two-core Intel® Atom™ processor E3827 running at 1.75 GHz, is designed for any environment and a multiplicity of sensor connectivity options. This connectivity flexibility is achieved using Dell-certified ISV partners, so options can range from wireless (ZigBee, 6LoWPA and Z-Wave) to wired (BACnet, Modbus and CAN bus).
The Intel® Atom™ processor E3800 series system-on-chip (SoC) helps meet the intent of the Edge Gateway, in that it’s relatively low power consumption allows the gateway to be fanless, thereby reducing noise and cost, while also enhancing reliability as the gateway has no moving parts. According to Intel, the processors deliver up to 3x the performance at one-fifth the power consumption of the previous generation. The horsepower drives many on-chip functions, including high-bandwidth interfaces such as PCI Express Gen 2.0, Hi-Speed USB 2.0, and USB 3.0, as well as integrated graphics support.
As these gateways evolve to be a nexus of data aggregation and analysis, security is being addressed using Intel® Advanced Encryption Standard New Instructions (Intel® AES-NI) which is built into the processors. This enables hardware-assisted data encryption and decryption for data on the move and at rest. Hardware support comes via the gateway’s Trusted Platform Module (TPM) IC that provides hardware root-of-trust, Secure Boot, and BIOS-level lockdown of unused I/O ports. Operating system support includes Wind River, Linux, Ubuntu Snappy Core and Microsoft Windows 10 IoT Enterprise.
Figure 2. Dell worked with companies such as KMC Control to let multi-tenant building dwellers control their environment and track their energy use using the KMC Commander on any smart device. (Source: KMC Controls)
Of course, Dell tried the gateway with some customers before its introduction. With ELM Energy the gateway proved it could do localized automation of energy transfer as demand from different sources fluctuated. With KMC Controls, the gateway formed the hub of the KMC Commander software that allowed multi-tenant building dwellers to control their environment and track their energy usage via a smartphone (or other intelligent system). It moved data to the cloud, while also enabling local analysis.
The IoT Analytics Gateways will greatly reduce latencies and enhance performance at the edge by perform data analysis and filtering out data that really needs to be sent to the cloud versus that which can be processed locally. Now localized events, from turning on a light to managing a factory floor, can be handled more quickly, effectively and securely. The trick, however, is determining what should be done locally and what data should be transferred. That’s a story for another time.