teknowlogy´s top IT trends for 2019 – 8/10 Edge computing
Edge computing supplements the centralized cloud computing approach with a decentralized component. While cloud computing provides all functions from a central data center, edge computing shifts data processing to the edge of an IoT installation.
Edge computing devices are available in every conceivable form, from small controllers, single-board computers, simple industrial PCs, and high-performance servers through to mini data centers. In any case, what is important is that computing power is provided as close as possible to the data source. Proximity to the data sources shortens the transmission paths to an entity that provides for initial processing or aggregation. This is important in environments where there is no network connection, or only a narrowband, low-performance one, or where real-time data processing is required. The latency that arises as the data transfers to the cloud or another central data center and back again is too big for this. Depending on the power of the processors, edge devices perform initial analysis tasks based on locally collected data, or consolidate, sort, and compress this information, resulting in less data traffic to the central data center.
With growing IoT penetration, increasing networking of devices and machines, and the proliferation of smart products, the need for edge computing is on the rise. An illustrative example is the self-driving car, which has a great deal of intelligence on board in order to respond to changing traffic situations in real time. In addition, there is a permanent connection to central data centers to obtain weather and traffic information, for example, or to transmit vehicle data. Further fields of application of edge computing arise, for example, in production, where machine data is preprocessed locally.
Edge computing will receive an extra boost from the next mobile communications standard, 5G. 5G is paving the way for a wider distribution of smart products, so that, for example, AI services such as voice and face recognition can be processed on-site and results can be aligned with a central entity.