When intelligence spreads from the cloud to the edge How to find gold mines?

www.techweb.com.cn/cloud/2019-02-11/2...

As more and more workloads appear in the cloud, and new technologies such as the Internet of Things enter production and life, additional demands are placed on local processing capabilities. Gartner once said in a report that edge computing is "eating" the cloud. The concept of Edge Computing is not new, but it is now appearing in more and more applications, like Microsoft is building a "smart shop" with American supermarket brand Kroger with smart cloud and edge technology. Renovated retail stores near their respective headquarters.

It is understood that the scope of cooperation between the two sides involves smart shelves, terminal equipment, store shopping efficiency, pricing, replenishment, and advertising conversion. Take "Digital Shelf System (EDGE)" as an example. It replaces the paper price tag with a digital shelf display and can display information such as product promotions.

Currently, the system has been deployed in dozens of Kroger chains in the United States. Compared to traditional shelves that require manual stickers, barcodes, etc., the new system running on the Azure platform can update 20,000 price information in minutes.

For example, if a consumer wants to buy a brand of pasta sauce, then when it comes to the shelf of the pasta sauce, the shelf system will send a signal to some dozens of sauce brands. In terms of the shelf, it is no doubt that it can improve the purchasing efficiency. In addition to EDGE, the retrofit system also includes a shopping guide experience, personalized advertising and more. Customers can generate shopping lists in advance through the Kroger app, which guides them through the store to purchase items.

According to a data presented by NetApp, the Asia-Pacific region is expected to have 8.6 billion Internet of Things (IoT) devices by 2020, and by 2025 will become the largest region of the world's 5G network, with 5G connections reaching 675 million. To take full advantage of the massive amounts of data generated, companies must have the ability to process data at the edge to gain insight and make real-time decisions. As a result, IoT devices and applications will increasingly include services such as data analysis and data reduction to make it more reasonable, faster, and smarter to decide which data needs to be processed immediately, which data needs to be sent back to the core or the cloud, or even What data can be discarded.

For manufacturing, much will benefit from smarter edge equipment. Using IoT devices to process data at the edges, manufacturers can perform predictive maintenance by detecting early indications of equipment failures, which helps prevent failures that hinder production or make unnecessary maintenance checks.

The automation system developed by Hollysys supports the daily operation of 25000 industrial systems around the world, covering China's high-speed railway, subway operation and nuclear power, thermal power plant automation control and other markets. It is not easy to manage and maintain more than 20,000 sets of system equipment in the world. According to the traditional way, it needs to consume a lot of manpower, material resources and energy to carry out on-site service, inspection and regular maintenance.

Rapidly growing businesses and growing customers have brought growing demand for on-site services, along with a growing range of services and investments, as well as a series of issues of management, communication, coordination, etc. . In addition, the traditional way of on-site service is inefficient. After the on-site technicians find difficult problems, they often need repeated technical communication with the headquarters, plus the time spent on equipment parts replacement, often requiring equipment to be shut down for maintenance.

For large-scale enterprises such as equipment production lines and power plants, production suspension means huge losses. Taking a boiler plant in a thermal power plant as an example, the maintenance itself may take only a few hours, but it takes a long time to wait for the boiler to cool down and reheat the water in the early stage. It must be stopped for at least one week before and after. The direct economic loss caused by the tens of millions of power plants has been quite amazing.

Obviously, avoiding unplanned outages to the maximum extent, improving the efficiency of operation and maintenance, and accelerating the speed of maintenance response are the core problems to be solved by Hutchison. This requires the help of cloud computing, Internet of things technology to create remote real-time inspection and diagnosis. Technology expert remote service, build a set of industrial automation control platform,

On the whole, the cloud computing service is mostly on the device side, and the Internet of things is the key bridge to connect these ends, whether it is the later generation of mobile edge computing or mobile cloud computing, which is leading the center of gravity to the end. Instead of wasting energy on the traditional IaaS or PaaS end. Therefore, cloud service providers will still follow the principle of "big intelligence" in the cloud and "small intelligence" on the edge when handling the workload, but in the long run, it has become an inevitable trend to build intelligent solutions around edge scenes in upstream and downstream industries.

According to Metcalfe's Law, the network value is proportional to the square of the number of users. When more and more people and smart objects are connected to a network, the entire network will be added. Research institutions predict that in the future, 79% of IoT traffic will be accessed through gateways, 50% of network traffic will come from the Internet of Things, and the Internet of Things will contribute more than 50 billion connections. At the same time, the amount of data generated by each person and every object will be 2000 times the current amount. This is a huge business opportunity, not only for pipeliners, but also for upstream chip vendors and service providers.

In addition, the consumer experience is dominated by edge computing, not the enterprise, which makes the cloud service provider's design logic need to be adjusted. For example, everyone knows that when the hand is about to touch the fire, it will quickly reach back and will not wait until it is burned and then react, which involves delays in response. Edge computing, as the name implies, is to complete the operation on the data source side. The device that completes the operation can be either a smart home appliance, a PC, a mobile phone, or even a camera, which can be realized through a local area network. At this point, edge computing also acts as a message dispatcher.

Of course, edge computing can only be more of a supplement to cloud computing, and it would be biased to replace it. Why? First, the evolution of IT complexity makes the isomerization data increase gradually, which puts forward the test of computing platform in processing performance. Second, how intelligent edge computing is, at least at present there is still a gap between the intelligent cloud and large-scale centralized data fusion of a variety of algorithms, for the application of multiple possibilities; Third, the industry standards for edge computing are still missing, the ECC alliance is just the beginning, and commercial applications still need to be explored.

Whether computing and processing data in the cloud or using artificial intelligence to train data, large-scale data processing is inseparable from the support of powerful computing power, which is also an important guarantee for accelerating industrial digital transformation and creating intelligent services. Nowadays, intelligentization at the edge has become the consensus of technology giants such as Intel, Microsoft and Cisco. However, in the actual scenes such as industry, energy and manufacturing, it is no small challenge to realize true "edge intelligence". . The author has interviewed Raejeanne Skillern, vice president of Intel Data Center Group and general manager of cloud service provider group, that the platform's strategic perspective, rich software and hardware tools, and open cooperation concepts are indispensable.

In fact, very few companies or solution providers can connect a section of a project from the cloud to the edge very well, which is very difficult. That's why Intel bought Movidius because the edge is not smart enough yet, and we've been investing in the edge. After the edge has processed a lot of things and then processed into the cloud, the cooperation between the cloud and the edge can reduce the bandwidth cost and demand, and push the solution closer to the customer level. Therefore, how to extend from the intelligent cloud to the smart edge is a problem that upstream and downstream manufacturers are thinking about. It is inseparable from the joint efforts of chip manufacturers, cloud service providers and solution providers. Only in this way can the edge computing dividend be realized. Maximum value.

Zhi can from cloud spread to edge Amoy gold mine

Read More Stories

© NVBOOK.com , New View Book