We have come a long way since Thomas Edison invented the electric light bulb and the industrial revolution democratized artificial power. Today, most of us take reliable, affordable power for granted. Historically, steady growth in demand for power allowed utilities to invest heavily in generation and delivery infrastructure and to maintain affordable prices for customers and reasonable returns. Increased efficiency, conservation efforts, and alternative power sources are changing the landscape, however. The industry is looking beyond traditional cost-of-service models and focusing on asset utilization and streamlining costs through operating efficiencies.
Internet of Things (IoT) applications offer possibilities for electric utilities to improve the efficiency and performance of the power grid: collecting data from sensors to measure the stability and operational efficiency of the grid and intelligently managing the grid based on advanced analytics. A key aspect of successful IoT applications is the ability to apply analytics at various points throughout the network and the data center, and to choose the type of analytics based on data volume, data velocity, latency and reporting requirements.
Such multi-phase analytics allow us to detect unusual events as they occur in the grid, proactively respond to an event that may have the potential of creating an outage scenario, while at the same time maintaining aggregated views across the grid and monitoring and updating operational models. This is just one example of a disruptive change to an industry when IoT Analytics are applied at multiple layers of the network.
As IoT implementations move from theory to reality, it is important to have credible partners who support your journey. Networking giant Cisco and analytics leader SAS are partnering to announce the SAS, Cisco IoT Analytics Platform, industry’s first validated architecture for edge-to-enterprise IoT analytics. This reference architecture combines the power of Cisco’s networking and data center infrastructure with SAS’s capabilities in streaming analytics and advanced analytics in a foundation for cross-industry IoT implementations.
Thanks to ubiquitous connectivity, millions of devices are producing streams of data about the physical world. Cisco estimates that 50 billion devices will be connected to the Internet by 2020, a number that is estimated to increase ten-fold by 2030. These devices communicate with each other, with network gateways, and with the cloud. A traditional architecture in which the data center or cloud is the only analytic execution environment to drive insights will not be able to handle the increasing data volumes and requirements for low-latency decisions. Analytics needs to move to the edge and to the aggregation points on the networks.
Cisco has made great progress in establishing compute platforms towards the edge of the network, such as IoT gateways that act as aggregators for IoT end-points. This gives us the ability to distribute analytics across the network and into the edge. However, edge analytics must be integrated into the complete analytics lifecycle, so that models created in the cloud or data center can be easily deployed and managed across the enterprise.
In our validation work with Cisco, we created a reference architecture for this scenario by choosing their 829 series of industrial routers as edge devices and unified compute system (UCS) servers as the data center representation:
To validate the reference architecture, we used sensor data from a smart grid containing millions of events. The data were ingested by the SAS Event Stream Processing (ESP) instance deployed inside the 829’s at high frequency. The analytic model executing inside the SAS instance detects unusual events on the grid. The results stream to the ESP instance in the data center, it acts as the aggregator of data from several edge sources. The ESP instance on the UCS provides a real-time “control center” view of unusual events happening on the edge end-points. To support deep advanced analytics and model building, the data are passed to SAS in-memory solutions on Hadoop. Improved models and additional analytic tasks can be deployed back out to the edge.
Whether it is a connected car, a smart grid, a smart factory, or a smart city, your IoT data are only valuable if you can act on the data in the right manner and in a timely fashion. Data without analytics is value not yet realized. Together, SAS and Cisco have created a validated design for edge-to-enterprise IoT Analytics. We are excited to help our customers making real-time decisions based on their IoT data.
Get the Cisco perspective on this validation project from Raghu Nambiar, Cisco Computing Systems Product Group CTO.