The battle between cloud computing and IoT platform opens the "second half", and the edge computing becomes the "main position"

Because of the cloud cake, Amazon, Microsoft, and Google have all tried their best to get rid of it. According to the data, Amazon, Microsoft and Alphabet (Google parent company) totaled $31.54 billion in capital expenditures and capital leases in 2016, a 22% increase from 2015. Each company lists cloud computing as a major investment area.

As the expansion of cloud computing is more and more radical, and almost into the "prisoner's dilemma." These three companies seem to have waged price wars, have reduced the price of some cloud products, and even caused concerns that prices may be too low due to continued price declines.

In this 1.0-mode universal cloud infrastructure service market led by Amazon, Microsoft and Google, is there no chance for latecomers? Not necessarily, the new gameplay of 2.0 mode cloud computing + edge computing is in the ascendant, cleverly becoming the turning point of cloud computing to avoid the "prisoner's dilemma."

Looking at the battle for the IoT platform, as the IoT is gradually affecting all aspects of production, manufacturing, and life, the IoT platform has become a strategic focus, and the number has soared in recent years. Whether it is Ali, Tencent, Baidu, Jingdong and other Internet giants; or Huawei, China Mobile, China Telecom, China Unicom and other operators, communications predators; or IBM, Microsoft, Cisco and other traditional IT companies; and Wisdom Cloud, Ablecloud, Qing New Internet of Things companies such as Branch and Bolian want to get a triangle in the big cake of IoT platform.

Although they are called IoT platform, the difference between connotation and substance is not small. If it can't be connected with the underlying equipment, the upper and the industrial application, and effectively empower the industry partners, the IoT platform can only be a false proposition.

Whether it is the cloud computing battle that needs to be upgraded, or the "red-eye" IoT platform campaign, the second half points to the same main position: edge computing.

Edge computing is not a reinventing concept

Peter Levine, a partner at Silicon Valley Ventures A16Z, once said that edge computing is the "terminator" of cloud computing. This statement is inevitably exaggerated for the eyeballs, but the reality of cloud computing and edge computing symbiosis has become the mainstream prejudgment of cloud computing and Internet of Things practitioners for the future.

From the technical definition, edge computing is an open platform that integrates network, computing, storage, and application core capabilities on the edge of the network near the source of data or data. It provides edge intelligent services to meet the industry's digitalization in agile connections and real-time services. Key requirements for data optimization, application intelligence, security and privacy protection.

Edge computing is not a new thing, nor is the IoT person boasting the concept of re-creation, but a counter-attack of the “long-standing” IoT distributed computing.

The characteristic of distributed computing is that each node has a computing function. The disadvantage is that each user needs to manage his own nodes, hardware, and software. So cloud computing appeared later, and a lot of data processing was given to the "cloud" to do it. This cloud computing is actually a centralized calculation, which solves the user's management troubles for central computing.

By the time of the "cloud", we have completed the transition from distributed computing to centralized computing. However, we now find that different IoT scenarios such as chicken feathers are not the best strategy by relying solely on centralized cloud computing. The “quick-calculation” capability of the edge is especially important for IoT applications.

The giant took the lead and turned to embrace "edge computing"

According to IDC's prediction, by 2019, 45% of the data created by IoT will be stored, processed, analyzed, and manipulated through edge computing. "Big things as wisdom" is the way to calculate the "accession to the WTO" by marginal computing. The leader of this counterattack is led by major giants and mainstream forces.

At the Microsoft Build 2017 Developer Conference this week, CEO Satya Nadella announced that Microsoft has met the new world: a world of Intelligent Cloud + Intelligent Edge.

Microsoft unveiled the Azure IoT Edge service, a cloud service for the Internet of Things, at the Build conference. It has sensors and small computing devices that track data in industrial scenarios and then analyze it by Microsoft's cloud and AI tools. This feature pushes computing power from the cloud to the edge.

These actions also indicate that computing power is going to the edge, because more and more IoT terminal data will require more computing power to sink, which also means more distributed AI and distributed computing.

Amazon, the world's largest cloud service provider, is also counting on the development of cloud platforms through IoT edge computing. At the recent AWSre: Invent conference, Amazon announced the launch of AWS Greengrass.

AWS Greengrass is software that allows users to perform local computing, messaging, and data caching for connected devices in a secure manner. With AWS Greengrass, connected devices can run AWS Lambda functions, synchronize device data, and securely communicate with other devices without even connecting to the Internet, minimizing the cost of transferring IoT data to the cloud.

In April of this year, the Linux Foundation released the open source IoT edge computing project: EdgeXFoundry. EdgeX Foundry is not a new standard, but a way to unify standards and edge applications. Its main purpose is to create and promote EdgeX, a universal open standard for the Internet of Things; around interoperable plug-and-play Components/components create an ecosystem; certify EdgeX components/components.

Also in April, network giant Cisco and business intelligence company SAS announced that it has developed the world's first edge computing platform for IoT analytics that combines Cisco's edge computing products with advanced analytics capabilities of SAS.

SAP has successively acquired two IoT edge computing start-ups: PLAT.ONE in Italy and Fedem Technology in Norway. After the acquisition, the SAPHANA cloud platform can support IoT business applications across the enterprise, both in the cloud and on the edge.

Dell is also actively expanding its edge computing platform products and industry applications, with additional release of edge computing products specifically for industrial applications. Recently, Dell's Edge Gateway 3000 series has real-time intelligent processing capabilities, occupying a small space and adapting to harsh environments, enriching the edge computing product lines such as Dell Edge Gateway 5000 Series and Embedded Box PC3000/5000 Series.

Huawei, the domestic communications giant, is also an active promoter of edge computing. It participated in the joint initiative to launch the Edge Computing Consortium (ECC).

Just this week, there is another development worthy of attention. The startup, Neurola, announced significant progress in deep learning, no need for a cloud server, and the ability to learn incremental objects at the edge. This means that artificial intelligence is also rapidly evolving from the cloud to the edge: autonomous vehicles can be personalized for each owner or specific area; parents can teach a toy to identify their children without fear of infringement Privacy; industrial-grade machines can be upgraded autonomously for specific tasks.

Complete the transition from cloud to edge with MDC

The real-time data generated by sensors is very different in form and function from traditional enterprise data. Whether it is gigabyte data generated from connected cars or control data from industrial robot assembly lines, edge calculation must be More agile, more autonomous, and more reliable than any cloud technology developed in the past.

Before using the Internet of Things system, many people did not realize that a large amount of IoT data may never be transmitted to the cloud, and it is only suitable for processing on the spot. If it is not processed in real time, the value of the data will be lost.

Some data have a very short “preservation period”. Once the treatment is delayed, it will quickly “degenerate” and the value of the data will fall off the cliff. So not all data must be uploaded to the cloud platform, not to mention the critical information may be delayed or interfered with during the transmission process, especially those transmitted over LPWAN (Low Power Wide Area Network).

We must respond quickly to these critical data to make decisions, either by taking action in a short period of time or by looking at the best time.

The rise of edge computing also means that the giants must abandon their dream of fully controlling or leading the IoT market, but instead focus on broad collaboration across multiple layers of data architecture. As an idea after an iterative upgrade, edge computing will redefine the relationship between clouds, pipes, and ends.

So what is the best way to advance edge computing for industry applications?

Some people try to do it themselves, explore new technologies, and hope to combine them into a viable solution. However, some people have found off-the-shelf, using edge computing solutions such as the Micro Modular Data Center (MDC).

MDC has been around for a while, and it is a set of hardware and software solutions for real-time acquisition, reporting and charting of detailed manufacturing data and processes in the shop floor. With the development of the Internet of Things, it is necessary to use a smaller MDC, that is, "micro-MDC", to quickly configure these modules to the edge scene to realize the deployment of computing power.

In the selection process of the micro MDC, the person who suggested came over to consider the following five points:

Ensuring flexibility: Due to the ever-changing variety of application scenarios, micro-MDCs need to consider enough flexibility. Integrated solutions are not necessarily the best choice. It is often necessary to perform micro-MDC optimization steps for different environments.

Open foundation: On the IT side, there is a need to find scalable and manageable solutions to distribute more processing power to the edge. This poses a great challenge for the IT department. In order to achieve service agility, with an open and agile underlying infrastructure, it is extremely critical to automatically scale resource allocations as needed.

Empowerment analysis: It's one thing to deal with edge data, and analyzing edge data is another matter. Micro MDCs should have the ability to perform in-place analysis, the closer the high-speed devices or sensors that generate the data, the better, so that business insights can be quickly delivered.

Fast use: Not only does the micro MDC have flexible options for computing, storage, networking, power and cooling, but it should also be offered in a quasi-integrated solution that will ensure rapid installation and rapid value creation of the miniature MDC.

Unified management: Most of the solutions for edge computing are scattered, and it is not easy to achieve unified management. To do this, device providers need comprehensive integration and management capabilities to manage micro-MDCs, related data centers, IT equipment, and sensors from around the world from the same portal.