Edge computing is a distributed computing framework, which derives cloud edge devices on the basis of traditional client / server (C/S architecture) or browser / server (B/SS architecture). "Cloud" is the central node formed after the scale of traditional server stack, and "end" is terminal equipment, such as mobile phones, smart appliances, various sensors, cameras, etc. With the rise of Internet of things (IOT), 5g, AI and robotics, the density and frequency of enterprise data production, consumption, processing and storage have increased by orders of magnitude. The traditional centralized architecture has been difficult to carry. Cloud computing capacity has sunk from the center to the edge. Edge computing energy efficiency will promote the formation of a collaborative computing system integrating "cloud, edge and end".
According to the principle of TCP / IP, the TCP sliding window is limited by the time delay value (RTT). In short, when the delay is too long, increasing the bandwidth will not help to solve the bottleneck of processing capacity. Therefore, if the time delay between data sources and computing facilities is too long, the investment in IT infrastructure will get half the result with half the effort, especially for the global networks of multinational enterprises.
The data of modern production is processed and extracted from massive metadata. Whether it is connected cars or intelligent robots in factory workshops, the amount of data generated by various devices around the world is much more than ever before. However, most of the data in the Internet of things have not been developed and utilized at all. For example, a McKinsey & Company study found that an offshore oil drilling platform generates data from 30000 sensors, but currently less than one percent of the data is used for decision-making. Excessive data accumulation also brings difficulties in data governance.
Edge computing provides a more efficient alternative: processing and analyzing data closer to where it was created. Since the data will not be transmitted to the cloud or data center through the network for processing, the delay is significantly reduced. Edge computing and mobile edge computing on 5G network support faster and more comprehensive data analysis, create opportunities for deeper insight, shorten response time and improve customer experience. Such a computing architecture close to the data source can bring significant business benefits: faster insight, shorter response time, improved bandwidth availability, and reduced expensive global infrastructure and network investment.
Edge computing originated from CDN (content distributed network). Based on the early content access model and bandwidth limitations of the global Internet, the buffer node set in the network greatly alleviates the bottleneck between the edge traffic and the central node. However, CDN only solves the problem of accelerating access to static content, but it has very limited effect on dynamic content, enterprise key traffic and increasingly personalized application types.
Moore's law promotes the continuous evolution of computing power of computing chips in the global x86 architecture, and urges infrastructure service providers to think about two important strategic issues: first, with the increase of computing power, how to reuse the spare computing power of devices on the side that are more receptive to the terminal and manage the deployment of workloads performing these types of operations? In fact, this problem has been solved by continuous iterative virtualization and multi cloud management technology. No matter the load balancing for the underlying link and computing resource overhead, or the controller monitoring the IOPs performance of each element, it has been able to integrate the computing power on the central side and the computing power on the edge side and develop in coordination. The cloud center focuses on large-scale, global and non real-time computing and analysis tasks, while the edge side focuses on small-scale, regional and real-time response and closed-loop transactions.
How to embed only intelligent algorithms in the device without increasing the cost of computing power, so as to affect employees, customers and business in a more responsive way? The evolution of network functional Virtualization and management, automation and network orchestration supports the realization of such an idea. Among them, network function Virtualization includes the basic network architecture components on the platform (router, firewall, load balancing, log system, etc.), and network orchestration provides a framework for managing infrastructure and preparing new vnf. Under the centralized architecture, the investment in network hardware construction is expensive, but when placed in the scenario of regional centers and even branches with relatively less investment, reusing computing power to make up for the shortcomings in management functions is a very cost-effective scheme.
Personal data has always been an indispensable tool for public operated enterprises. However, the potential management risks in the process stage from collection, processing to utilization of personal data continue to increase, coupled with frequent data leakage events, which highlights the looseness of personal data management in the past.
With the formal adoption of the general data protection regulation by the European Conference in April 2016, it will be officially implemented on May 25, 2018. In fact, the above regulations require more records to be retained and demonstrable process methods to be adopted, rather than just staying in the past and "no violations have actually occurred". Edge computing is undoubtedly a practical answer at the infrastructure level to help ensure privacy and comply with local laws and regulations on data storage and retention.
At the Annual Global Climate Summit held in Glasgow, Scotland in 2021, global leaders made commitments to reduce carbon emissions and signed historic agreements. Any multinational enterprise committed to the global market has the responsibility to face the goal of environmental protection and establish corresponding values to obtain the recognition of users. Large data centers have a lot of criticism on energy consumption, refrigeration and thermal effect management every year. Therefore, edge computing actually apportions the load of large cloud computing centers, promotes the realization of Green IT in business model, and gives enterprises the label of environmental protection and energy conservation.
Although this process is not without challenges, an effective edge computing model should be able to solve the network security risks, management complexity, delay and bandwidth constraints. A viable model should help you achieve the following benefits:
1. Present global real-time performance changes to users with visual data reports.
2. Manage the workload on all cloud environments and any number of devices.
3. Reliably and seamlessly deploy applications to all edge locations.
4. Remain open and flexible to meet changing needs.
5. Operate more safely and confidently.
