Edge Computing is one of the latest buzz expressions in the ICT world, along with IoT, big data, edge analytics and cognitive computing.
What is Edge Computing?
Edge computing refers to providing data processing power at the edge of a network, near where the data is produced and collected, as opposed to transferring the raw data to the cloud, or a central data warehouse, for processing. "So what?," you may well ask. "Why do we need yet more ways for processing and storing data?" or "Why not let all the sensors send their data over the network to a cloud?" The fact is that a lot of data is produced at the edge, and this trend is predicted to massively accelerate. Further, even if you could send all of this data reliably to a cloud, in many cases this data loses its value the longer it takes to be processed. In some cases this is not important, but often it is critical. If you want to know whether someone is hacking a critical infrastructure, or want to adjust light intensity to improve the safety of motorists, you need the most up-to-date information. Waiting for the data to make its way back to a cloud or data centre for analysis is likely not to be good enough or safe enough.
Edge computing can take the form of micro data centres or small machines placed in remote locations, or at the edge of the network, to provide real-time insights into data. It then acts as an aggregator or a gateway when operated in conjunction with wireless sensor networks, high-definition audio or video equipment and so on.
In a nutshell, edge computing enables the transformation of data into information, at the edge. The benefits are considerable:
- Access all of the data, without network limitation
- Provide a new line of defense against cyber attacks
- Improve responsiveness of systems, taking advantage of local decision-making capabilities
- Anonymise data during the transformation process, avoiding data privacy issues
- Massively reduce network usage, since only processed and valuable information is transferred from the edge to the cloud and back
- Reduce costs in operations and centralise resource needs
Now let's look at a few concrete applications of edge computing.
Examples of Edge Computing
Next generation systems, embedded with high density connected components, promise to be an enormous source of data. For example, the C-Series, the latest airplane by Bombardier, could be producing over 844 TB of data per 12 hours flight. Clearly, airlines operating fleets of this type of craft will be faced with a deluge of data.
It is not realistic to expect to transfer, store and process this much data. Nor does it make sense to do so. Analysis applications need structured information instead, in order to predict important events and derive significant trends. This information, using edge computing, could be produced from the raw data, closer to the components or system generating the data. In this example of the C-Series, the aircraft itself, is probably the right place to perform this type of data transformation.
Another example of smart use of data at the edge is the ability to control street lighting based on traffic level. Here, light intensity is adjusted in real time, based on the volume of traffic passing. This not only ensures safety for motorists and pedestrians, it also offers a huge opportunity to municipalities and transport ministries to reduce energy bills and meet environmental targets. If you would like to read more about this example, take a look at our NuvlaBox page.
As for the near future, SixSq is currently part of the mF2C H2020 project, which aims to deliver the ability to feedback high-level analysis down to the edge layers to optimise the transformation of data into information. This would potentially address the risk of losing relevant data, in the case where the transformation engine was not optimally calibrated. Take a look at the mF2C website for some interesting real-world use cases, like emergency situation management.
A Word on Automation
If we project edge computing deployments to the scale that analysts generally predict, the challenge shifts to operations. How do we keep so many smart connected device up-to-date? How do we monitor and control these fleets of devices and keep them secure?
This is where automation becomes key. Whether you use edge solutions emerging from network or cloud experts, remote management and devops capabilities should be on your checklist. This will ensure the edge is able to evolve with the rest of the systems, avoiding silos and ensuring cost control in terms of operations.
Edge Computing at SixSq
Coming from a cloud computing background, we have long realised that the benefits of cloud computing could be miniaturised into a platform now called edge computing. Our NuvlaBox is a hyperconverged solution delivering a generic solution to implement edge computing. The example above on smart lighting is our latest fully integrated solution developed in partnership with Schréder. Our direction of travel will lead us to more miniaturisation to enable further smart city and Industry 4.0. On the big data side, cognitive computing now embodies the fusion of machine learning and edge computing. More on this soon.
If you have any comments or questions about this topic, don't hesitate to get in touch with the SixSq team.