This article introduces the origin of edge computing and the relationship between cloud computing, and comprehensively sorts out its value and application scenarios. The full text is about 2200 words, and the reading time is 7 minutes.
A single technology cannot meet complex and changing user needs, which is especially evident today, with typical examples such as cloud computing and edge computing.
Cloud computing has long been expected, and with the continuous development of services such as video cloud and 5G, edge computing has now proven its value. So in 2019, edge computing ushered in a period of rapid development, and the popularity continued to increase. In the view of Baidu Smart Cloud, 2021 will be the first year of the outbreak of edge computing, and by 2022, 40% of computing tasks will be completed on the edge.
What exactly is edge computing? Is it something new? Why is the limelight so strong in just a few years? What value can edge computing bring? What are the advantages? Who are the players? Who is in the lead? This article makes a comprehensive review.
〓Edge computing glows in the new year〓
As early as 2003, Akamai proposed the concept of "edge computing" in cooperation with IBM and developed some initial applications. Therefore, edge computing is not a new term, but it has been revived in a specific era. In the technology circle, such things are not uncommon. For example, artificial intelligence is only able to truly land after two incidents and two lags behind.
Why is edge computing breaking out again? The fundamental reason is that the popularity of the Internet of Things leads to exponential growth of data, making cloud computing not the optimal solution. IDC predicts that by 2020, there will be approximately 50 billion smart devices connected to the network worldwide, mainly including smartphones, wearable devices, and personal vehicles, among which 40% of data needs to be processed at the edge.
It is not difficult to understand why edge computing has become so popular in a short period of time. And there is another set of data that is more noteworthy. According to the data released by IDC Global DataSphere in November 2018: more than half of the world's data was created in the past two years, and this trend will continue.
In other words, the strong demand for edge computing is not a flash in the pan, but will become more and more prosperous.
〓Value comparable to cloud computing〓
What problems can edge computing solve? Through a vivid example, edge computing can be understood in a popular way, that is, octopus.
Octopuses are known as one of the most "intelligent" biological groups on earth. In April 2016, an octopus named "Inky" in the National Aquarium of New Zealand climbed out of a half-open aquarium, walked across the room and drilled into a drain, and after passing through a 50-meter-long water pipe, successfully escaped into the open sea. One of the most powerful proofs.
The reason why an octopus is smart is that it has "one brain + multiple cerebellums". It can not only analyze and make decisions through 40% of the brain capacity, but also perform perception and analysis through 60% of the huge number of neurons distributed on its eight legs.
Edge computing is like the octopus's cerebellum, and cloud computing is the brain. The value of edge computing is obvious. By pre-calculating, the data can be processed locally, and those that need further processing are transmitted to the cloud, and those that are not needed are directly fed back. This results in faster response times and higher computational efficiency.
〓Just needed in many scenes〓
Because edge computing has a unique value different from cloud computing, it has become a basic requirement in many application scenarios.
In video scenarios (live broadcast, on-demand, etc.), before the emergence of edge computing, the video will be transmitted to the IDC source station for processing. There are two major problems here. One is that the overall upload, processing, and distribution links are too long and will freeze. In delay-sensitive scenarios such as live broadcast, the user experience will be poor. In addition, the bandwidth cost of IDC is relatively high, especially in such high-traffic scenarios, which will bring great pressure to enterprises.
After the emergence of edge computing, many video manufacturers began to upgrade the architecture of edge source stations. By forwarding the source station and video-related processing tasks to edge nodes for processing, including video stream access, review, merging, slicing, transcoding, or other video-derived computing tasks, such as barrage processing, etc. After processing, the distribution capacity of the original CDN is used to distribute the content to viewers across the country. Under this new framework, it has a very obvious effect of reducing costs and increasing efficiency.
At the same time, if the anchor and the client are in the same area, the effect of the distributed source station will be more obvious, just like local tasks are processed locally and distributed to the nearest local user side for viewing.
In smart security scenarios, there are various types of cameras, and the cost of adaptation is high. The unified access of video streams and picture streams through edge nodes can realize unified management of terminals. The end is responsible for part of the preprocessing of the video stream, and the processed video stream is uniformly aggregated to the edge node, and related structured processing, such as face analysis, vehicle analysis, behavior analysis, etc., is also placed on the edge for calculation.
If there is an interaction with the terminal, the result can also be returned to the terminal in real time. Other core data can also be sent back to the central cloud for storage. This is actually a good reflection of the three-body collaboration of cloud, edge, and end, and the computing power is digested layer by layer.
In addition, edge computing also has outstanding performance in industrial Internet, new retail, autonomous driving and other industries.
In general, with the rapid implementation of new technologies such as 5G, many application scenarios that were unimaginable in the past have been implemented, including virtual reality and smart home. These applications are driving the rapid maturity of edge computing.
〓Manufacturers are making efforts〓
As the importance of edge computing becomes more prominent and there are more and more application scenarios, the industry pays more and more attention to edge computing. In the past two years, we have seen that various manufacturers no longer simply promote cloud computing, but both, and even tilt more resources to the edge.
In a large dimension, edge computing mainly includes several major types of manufacturers: first, cloud computing manufacturers, including Baidu Smart Cloud, etc.; second, hardware manufacturers; third, CDN manufacturers; fourth, operators.
Each type of vendor has its own focus and advantages. For example, cloud vendors pay more attention to cloud-edge collaboration, hardware vendors focus more on selling equipment, and CDNs and operators focus on data center resources.
Based on their respective advantages, the services they can provide and the application scenarios they target are also different. Some simple scenarios may be met by various manufacturers, but some scenarios with a particularly large amount of data and high latency requirements, such as autonomous driving, can only be satisfied by a few manufacturers. Therefore, comprehensive consideration must be given when selecting an edge computing service provider.
〓Edge Layout of Baidu Smart Cloud〓
As a leader in the field of cloud computing, Baidu Smart Cloud has launched its own edge computing solution very early, and it has been continuously improved over time.
As of now, Baidu Smart Cloud has a complete edge product system layout, including cloud edge, MEC edge, and terminal edge. The nodes have completed global coverage, supporting many typical application scenarios of edge computing such as pan-video, smart security/city, cloud games/mobile phones/desktops, and dial-up testing platforms.
Among them, Baidu Smart Cloud is in a leading position in many segments. For example, in the AI edge computing industry, in the Akraino Edge Stack Release 3 officially released by LF Edge, an international open source organization under the Linux Foundation, the IME AI Edge Blueprint contributed and led by Baidu Smart Cloud successfully entered the Release 3 version and will soon become the core project of the Akraino Edge Stack Release 3.
The edge vision of Baidu Smart Cloud is Compute anywhere, to be the most agile and intelligent edge computing platform. For this goal, Baidu Smart Cloud is still relying on its advantages in the field of cloud computing to continuously increase edge computing.
To sum up the full text, the prospect of edge computing is unquestionable, but this does not mean that edge computing will completely replace cloud computing. It can only be said that both have their own strengths. Only division of labor and cloud-edge collaboration can support enterprises to meet the upcoming challenges of various data processing. Among them, Baidu Smart Cloud has a very comprehensive product layout and implemented many cases by virtue of its long-term technology accumulation. This is an advantage and capability that many similar manufacturers do not have.
Welcome to click [Read the original text] to learn more about Baidu Smart Cloud edge computing.
share, comment, like, watching
Come to "One Key Four Links"!