Every new information technology, from its inception to maturity, seems to be a process of feeling one’s way across a river by stepping on stones, but there are inherent rules to follow. As a bridge connecting 5G, artificial intelligence, cloud computing, and the Internet of Things, edge computing stands out as a leading emerging technology trend of 2020. According to IDC analysis, over 50% of data will rely on the maturity of edge computing for analysis and computation, with the global edge computing market expected to reach $250.6 billion by 2024. With high expectations placed on edge computing, it has gradually begun to reveal its future development path after initial explorations. At the recently held “2020 Edge Computing Industry Summit (ECIS 2020)”, the Edge Computing Industry Alliance (ECC) once again emphasized the importance of collaborative development between edge and cloud, promoting cooperation between industry and academia on theoretical innovations in edge-cloud synergy, paving the way for industrial prosperity and application implementation.
From Concept to Practice: Edge-Cloud Synergy Enters the 2.0 Era
Edge computing is thriving as it meets the demand for providing edge intelligent services nearby, driving the computational model from centralized cloud computing to a more distributed edge computing approach. Both cloud and edge have their strengths, and only through close collaboration between edge computing and cloud computing can better support various businesses, thereby amplifying the value of both.
“In 2018, ECC, in collaboration with AII, launched the ‘White Paper on Edge Computing and Cloud Computing Synergy’, which played a significant role in building consensus across the industry,” said Huang Haiqing, chairman of the ECC Demand and General Group, in an interview with China Electronics News. “Now, almost every company involved in cloud computing and edge computing is discussing cloud-edge synergy; this concept has taken root. However, in the practical implementation process, what challenges do we face, what kind of architecture is needed, and what key technologies are involved? These questions remain relatively vague, which is not conducive to industrial practice. This year, we are launching ‘White Paper on Edge Computing and Cloud Computing Synergy 2.0’, hoping to promote the practical implementation of edge-cloud synergy based on consensus.”
ECC Edge-Cloud Synergy 2.0 project manager Ruan Binfeng stated: “We have collaborated with 15 partners, including the China Academy of Information and Communications Technology, Haier, Tencent, Huawei, and the three major telecom operators, to discuss from the 2.0 perspective. We believe that edge-cloud synergy 2.0 has three main collaborative directions: application synergy, service synergy, and resource synergy. Each layer presents different challenges.”
The goal of application synergy is to achieve consistent experiences across distributed deployment points of edge applications. For the practical implementation of edge computing, application synergy is the core of the entire system, involving cloud, edge, management, and terminal aspects.
“In application synergy, we see many challenges. The first challenge is how to manage a large number of edge nodes relative to the centralized cloud, which is indeed a significant challenge. Currently, the industry has proposed many technologies to address these issues, such as P2P technology and buffering technology,” Ruan Binfeng said. “The second challenge is that there are applications on both the cloud and edge, such as smart cities and smart transportation, which are typical scenarios. How to achieve synergy between cloud and edge applications, and how to manage and distribute them uniformly; the third challenge is operational stability. There will be many issues in edge scenarios, with network and edge facilities being unstable. In such cases, how to ensure the continuity and reliability of edge-cloud applications is also under further exploration by ECC.”
The goal of service synergy is to enhance the efficiency of edge applications. Service synergy mainly includes two aspects: one is the capabilities provided by central cloud services and cloud ecosystem partners, including data, intelligence, and application enabling capabilities; the second is to provide a complete process for the access, discovery, use, and operation of edge services through a cloud-native architecture.
“Currently, we can see that the various services provided by different companies on the cloud, such as AI services, data services, and application services, have different standards, which can lead to fragmentation across industries. ECC hopes to promote the standardization of services and facilitate standardized access through an industry-standard framework,” Ruan Binfeng said. “A significant part of service synergy is intelligent services, which require a large number of samples. How to enhance incremental knowledge, achieve collaborative reasoning, ensure security and privacy, and how to store the vast amounts of data obtained in edge computing all present challenges.”
Achieving resource synergy can not only simplify the difficulty of application development but also improve the efficiency of edge resource utilization. “Resource synergy can apply some ideas and thoughts from cloud computing; however, in edge computing, there will be more connotations. In addition to coordinating computing, storage, and network, it will also involve traffic synergy. For example, how to achieve global scheduling, how to accelerate across the entire domain, and how to optimize the underlying computing and storage network resources, considering these three dimensions to address resource synergy issues,” Ruan Binfeng said.
From Industry to Academia: Edge Computing Calls for Theoretical Innovation
“As edge computing develops, it will inevitably produce a new technical system, just as the development of cloud computing has produced a new technical system. Therefore, collaboration between academia and industry is necessary to overcome the challenges encountered in the practical implementation of the industry,” Huang Haiqing stated.
At this edge computing industry summit, Chinese Academy of Sciences academician Yao Jianquan stated: “The intelligence of the digital world is evolving from centralized monolithic computing to distributed networked computing, and is progressing towards heterogeneous, collaborative, and fully ubiquitous intelligent computing. Edge computing is an inevitable trend in the evolution of computing architecture. This is akin to the evolutionary process in biological evolution, from single-celled organisms to multicellular organisms, and then to complex life forms. The collaborative computing architecture of cloud, edge, and terminal is similar to the architecture of the human brain, spinal cord, and peripheral nervous system.”
Yao Jianquan further discussed, “Edge computing should consider several theoretical issues: first, computability theory, which needs to consider different computing modes based on the characteristics of computing tasks and the feasibility of multi-mode collaborative computing; second, fuzzy computing theory, which needs to consider the scale, granularity, fuzzy objectives, and constraints of fuzzy computing; third, complexity theory, which needs to ensure the controllability of computing tasks and overall engineering in complex scenarios with multiple consumptions, constraints, and heterogeneous dynamics; fourth, efficiency theory, which needs to optimize performance from the perspectives of efficient data caching, network collaborative transmission, and elastic training inference.”
Yao Jianquan also proposed that in the infrastructure, platform architecture, and application services of edge computing, it is necessary to continue exploring the supporting theories and the constraints between various levels. In terms of infrastructure, it is necessary to integrate massive heterogeneous devices to provide resource support, considering the constraints of scalability, heterogeneity, and collaboration; in terms of platform architecture, it is necessary to provide the capability to bridge devices and services, considering the constraints of high availability, flexibility, and universality; in terms of application services, it is necessary to allocate service resources and adjust operational capabilities, considering the constraints of adaptability, isolation, and agility. “In summary, these issues are still not very mature, and we hope that academia and industry can sit down together to conduct theoretical research to provide guidance for future technological innovation and industrial practice,” Yao Jianquan said.
“Since the advent of edge computing, we have a new opportunity to ‘negate’ the overly centralized cloud service model and decentralize data,” said Guo Deke, deputy director of the CCF Distributed Computing and Systems Committee, in an interview with China Electronics News. “Cloud computing adheres to the basic theories and methods of distributed computing, and edge computing has more distinct characteristics of node decentralization than cloud computing. We must consider how to utilize the theories and methods of distributed computing in a decentralized manner to meet user needs for the convenience and speed provided by edge computing while ensuring consistency, reliability, and overall performance of cloud-edge synergy.”
From Centralization to Decentralization: A New Computing Pattern Will Foster Unicorn Enterprises
“Currently, the cloud computing market is highly monopolized, with a few large cloud computing companies becoming the leading enterprises, occupying the vast majority of market share,” Guo Deke told reporters. “The survival space for smaller cloud companies will become increasingly limited.”
In the trend of edge-cloud synergy, will future edge computing also become a feast for giants? Is there still room for emerging enterprises to grow?
“In the field of edge computing, unicorn enterprises will undoubtedly emerge,” Huang Haiqing believes. “The reason for this is twofold: on one hand, we see that the market space for edge computing is large enough; consulting firms estimate that the market space for edge computing will reach as high as $500 billion in the next five years; on the other hand, we can refer to the development history of cloud services, from the emergence of the cloud concept to its true industrialization, and finally to the emergence of service models and business models that have a significant impact on the socio-economy, which took about 15 years. Objectively speaking, the large unicorn companies or commercially successful companies that emerged around the cloud have only been around for the past five years. Edge computing truly began to rise around 2015-2016. Considering the development cycle of edge computing, it is still in the stage of partial industry practice in certain scenarios, but the large-scale application of cloud computing has laid a good technical, ecological, and model foundation for the development of edge computing, allowing edge computing to scale up faster than cloud computing. It is expected that in the next 2-3 years, the technical system surrounding edge computing will become increasingly complete and support the large-scale application of edge computing.”
Ruan Binfeng stated that from several levels of edge computing, it is relatively difficult for emerging enterprises to grow to the platform level, but there are many opportunities emerging at the application layer of edge computing, covering a wide range of scenarios.
“ECC has always emphasized that edge computing has six major value industries, including telecommunications, industrial manufacturing, power energy, intelligent transportation, smart cities, and digital entertainment,” Huang Haiqing said. “We believe that in the next 2-3 years, edge computing will first scale up in these industries. In this process of scaling up, unicorn-shaped enterprises may emerge more quickly.”