MARCH 2019CIOAPPLICATIONSEUROPE.COM8oT devices are becoming more and more intelligent and can now create meshed networks by themselves, switching from a sensor into an actor and transferring information solely for meshed neighbors. For example, a connected car could tell a future home that the homeowner will be at home in five minutes and the garage door and the door need to be unlocked in time, the lights need to be switched on, and the grid operator needs to be informed that the wallbox is now charging at 22 kV. In the near future, this will happen over direct meshed information cells, operated by connected devices, wearables, sensors, actors and mobile devices ­ in short: everything. And all cloud providers offer dozens of solutions to master the challenges in a number of different ways. Self-organizing mesh networking and communication comes with a permanent flow of information, with massive IoT data streams; even classic Big Data frameworks such as Hadoop cannot handle this in a timely manner anymore. Coming along with the art of data, the need for data processing changes with the kind of data creation and ingestion. Most analyses will be done on the edge and during the ingestion stream when the data comes to rest. The data lake should be the central core to store data, but the data needs to be categorized and catalogued together with a proper, well-defined scheme and data description. The intended use of gravity generates needs to be applied as the motor of data-driven innovation.Why? Batched processing helps predict value out of stored data while analyzing multiple other data points and storage facilities, but not to react in time. And timely information in IoT enables business processes to have valuable meaning at the time they occur. To do the job, stream processing frameworks such as Spark or Kafka are more suitable. Combining both techniques brings unmatched value and impact to the business, driven by the right use of data. Stream processing during data transportation closes the gap between rapid data and data on rest. Mostly referring to the more costly IoT at edge computing, MQTT-enabled stream processing engines deliver high throughput over all kinds of compute instances, be it in a local data center, hybrid clouds or in public clouds. The same is countable for available cloud technology. Every cloud provider has its own IoT solution zoo with its own lock-ins, but often they do not fit in with scaling plans either in complexity, missing or not well-implemented parts, or simply because the price model is not comparable to the margin from an IoT-based product. A combined approach of scalable cloud technology (which fits most) and own developments brings the most benefit at an affordable price tag, without mentioning the intellectual property a business gains Enabling IoT to Establish a Sustainable Value ChainALEXANDER ALTEN-LORENZ, CHIEF ARCHITECT, DIGITAL DEVELOPMENT & TECHNOLOGY, E.ON SEIAlexander Alten-LorenzIN MYOPINION
< Page 7 | Page 9 >