Some companies are already using the edge computing, by moving the processing of data at the end of the network. What could change this technology, the current model of computing?

The edge compunting is shown as a plausible solution to solve the problems of capacity and latency, which can generate the huge growth of data traffic in the coming years.

In the previous installment of this story explained in what consists the edge computing, and we cited some of the advantages it brings, as well as the difficulties that this technology can be found in your deployment.

what But what it does and how it is being used since the edge computing? “We have references in several fields of application. An interesting case is the use of these technologies in the design of smart buildings. Another case is relevant is that of a company that manages networks of wind turbines. Each stop time is not provided for in these generators carries penalties of operators of the same with the companies that market the power generated.

Through the use of techniques of machine learning -at the end of the network — and deep learning-in the data center-, the company in question is able to avoid these penalties, improving the profitability of the investment in this project”, explains Ignacio Arrieta, pre-sales manager for commercial Spain & Portugal, Dell and EMC.

Carlos Clerencia, country manager of Intel Iberia, points out that “we are seeing many improvements in the retail where it analyzes the behavior of customers inside the store, interacting with them through digital signage and mobile applications, achieving a higher level of loyalty. Also is doing a lot of work in smart cities which covers aspects such as traffic regulation, energy saving, public security, to name a few. The industrial sector is also getting great benefits with the automation of factories.

Also, Jorge Jimenez, product manager Iberia of IT Division of Schneider Electric, points out that “there are many organizations that employ this type of technologies, most retailers, such as fashion stores or fast-food businesses, who already have small solutions in their different locations. Many industries also have micro data centers distributed by your plants to ensure the production”. However, he notes that in many companies, it still it is necessary “to compact and to standardize such solutions to guarantee the maximum security and availability, but also to simplify the ecosystem and, therefore, the management of the problems.”

For its part, Siavash Alamouti, CEO of mimik, indicates that “we are applying some basic forms of edge computing. For example, the majority of applications are performing more and more functions in devices edge. The data storage on Facebook is a good example. There are also some vertical applications doing this, such as some apps of conversation peer-to-peer or shared storage. The next step is to extend this concept further and make the node edge to communicate directly, collaborate and share results. This has been our focus in mimik. We have shared our SDK with some partners and very soon we will see applications that use edge cloud distributed”.

Change of model?

Bringing computing to the end of the network can lead to a change in data traffic and the current computing model. «As telecommunications providers prepare for the IoT explosion, they will need to increase their infrastructure. Not only in terms of access and transport, but also in terms of How they connect to the cloud. Future demand will be immense and bandwidth efficiency and latency will be a great challenge. History has shown that Moore’s «law» on computing is at least twice as fast in the case of bandwidth. Thus, even if the number of devices does not vary, the networks will always be a bottleneck. If the cloud architecture stays as it is and the edge devices have to send all their data to the central cloud, the networks will be saturated and the existing infrastructure will not be able to scale, ” comments the mimik CEO .

Transferring data processing to the end of the network could detract from the ownership of the main cloud computing platforms, diverting part of the business to the companies in charge of controlling and managing a set of IoT devices. < / strong> In this scenario, telecommunications operators take a good part of the cake. “Different IoT device management platforms are emerging. The devices need to be managed and the data collected needs to be analyzed. And this requires a coordination of the cloud with the network. The operators have a huge opportunity with the arrival of 5G to position themselves in the field of the network. It remains to be seen who will be positioned as an IoT cloud platform, ”says Clerencia.

But a diametrically opposite scenario could also be contemplated. Thus, Arrieta indicates that there are those who predict a future in which “a ‘ commoditization ’of the transport layer occurs as there are better networks and several operators”.

In addition, Alamouti recalls that “leading cloud providers, such as Amazon or Microsoft, have already understood the importance of edge computing and have started sending new edge nodes to their customers, such as Amazon Echo, Google Home, etc. ”. In this sense, review that mimik offers the possibility of converting“ almost any type of device ”into an edge cloud node. «Not only are we not going against anyone, but we allow all ecosystem participants to make use of existing devices such as edge cloud nodes. For example, both Microsoft Azure IoT and Amazon Greengrass They can be hosted as micro-services on an edge node of ours without the need for additional hardware. We make it possible to connect ‘IoT islands’ to mimik edge clouds (clusters), allowing the creation of specific applications and use cases in which the IoT network overlaps physics. ”

In short, the product manager of Schneider Electric believes that “what is clear is that edge computing is necessary. And when IoT applications really start up like the autonomous car or virtual reality or increased, they will be more ”. On the other hand, he considers that “the form he will take will depend mainly on two factors: connectivity and energy. One must guarantee latencies and bandwidths. The other is an increasingly high cost in our country, in addition to being necessary for everything to work. ” Finally, he points out that “edge computing has arrived to stay and we believe that the form it will take in most cases will be with micro data centers ”.