Abstract : Today, the OpenStack Foundation officially released the white paper "Cloud Edge Computing: Beyond the Data Center" in Chinese. The Chinese name: "Edge Computing - Crossing Traditional Data Centers". Several community volunteers from Intel and Kyushu Cloud participated in the Chinese language. Version translation. The official release of the Chinese version of the Edge Computing White Paper not only represents the initial results of the OpenStack community in the field of edge computing, but also indicates that China will be an important strategic market for the future OpenStack community to transform and upgrade to OpenInfra.
In the past decade, the centralized cloud computing model has become a standard IT service platform. While cloud computing is now a common application, the evolution of the Internet of Things and the upcoming introduction of more immersive and interactive application trends will shift the focus of data storage and computing away from the central data center and to the edge.
New technologies, new demands, IoT outbreaks and more real-time applications will fundamentally change the computing topology in the next four to five years. As data and content move from centralized clouds and data centers to the edge, the IT infrastructure will be reversed, creating new computing and storage usage scenarios. The interaction between people and numbers will also become more authentic, interactive and dynamic. Digital business will blur the line between the digital and the real world, and "fuzziness" will mainly occur at the edge.
Today more and more companies, organizations and research organizations are turning their focus on cloud computing to focus on edge computing and moving the architecture of applications and data from the cloud to the edge. The location and distribution of computing and data storage plays a crucial role in the soundness of IT applications and management.
Faced with new demands and the emergence of new technologies, the OpenStack community set up a dedicated Edge Computing Working Group to pay special attention to the evolution of cloud computing to the edge. Following the English white paper, the Chinese version of Edge Computing was re-updated and officially released on April 11. White Paper, Edge Computing - Across Traditional Data Centers (hereafter referred to as the "White Paper").
What is edge computing?
In fact, the so-called "edge" technology is not a new technology. The so-called "freshness" is the change that occurs at the "edge". Compared with the way in which data was only collected and collected as the central end office of the data, the “edge” gradually became the main source and key node for dealing with explosive data, and more importantly, the interconnection with “people”.
The white paper clearly illustrates the concept of "edge computing", which provides application developers and service providers with cloud services and IT environment services on the edge of the network. The goal of edge computing is to provide compute, storage, and network bandwidth close to data inputs or users. An edge computing environment typically has the following characteristics: potentially high latency between multiple sites, network unreliability, and slow bandwidth, along with other delivery services and application functions that the general data center centralized resource pool cannot handle. By migrating some or all of the processing programs close to users or data collection points, edge computing can greatly reduce the impact on applications under large-scale distributed sites.
The "edge" of edge computing refers to being at the edge of the administrative domain, as close as possible to the data source or user. The same concept applies to carrier networks, large enterprises with many branches, such as retail, or other applications related to IoT. The concept of edge computing must include edge sites (such as compute, network, and storage infrastructure), and running applications (computing tasks). Applications in an edge computing environment can enjoy various resources in cloud computing, such as computing, block storage, object storage, virtual networks, bare metal, or containers.
For the characteristics of edge calculations, use cases and application scenarios, the white paper has also been deeply analyzed. The driving force behind driving edge computing is to be able to serve users or terminal data resources one step closer. The edge computing environment will work with the core to provide a better user experience without making unreasonable connection requests to the core. And it will improve by reducing the delay and reducing the bandwidth limitation to solve the bottleneck of response and bring more effective and real interaction and experience.
At the same time, the white paper also raises some of the problems that may exist. For example, in order to apply edge computing, it is necessary to increase the number of deployments. This has led us to the challenge of deploying on a large scale. There are still necessary trade-offs in the future.
Use cases for edge calculation
In this edition of the white paper, several important examples are highlighted. These use cases are the more representative application opportunities for edge computing and the focus of recent work. In summary, analytics, compliance, security, and network function virtualization are the four main types of work that benefit from distributed architecture:
1, data collection and analysis
The Internet of Things, usually with its data collected on a large-scale microsite network, is a typical application that benefits from the edge computing model. Often large amounts of data are transmitted over a limited network connection to an analysis engine located in a centralized data center for data analysis, which is counterproductive; the analysis engine may be under-responsive, potentially causing additional latency and wasting valuable bandwidth. Since edge computing devices can also generate tera-scale data, analyze the data close to the source for analysis, and send only small batches of summary information to the central system, which makes the analysis closer to the edge-side data source more cost-effective. Of course, this also requires a trade-off between the cost of delivering data to the hub and missing information.
Unfortunately, with the popularity of edge devices (including mobile phones and IoT sensors), the proliferation of terminal devices has also led to the emergence of emerging attack vectors. Edge computing enables security components to be closer to attack sources, launching more efficient security applications and increasing the number of tiers to protect against core infringement and risk.
3. Compliance requirements
Compliance covers a wide range of needs, including geofencing, data sovereignty and copyright enforcement. Data usage is constrained based on geographic and political boundaries, data is restricted based on copyright restrictions, and data is stored in areas containing specific regulations that are achievable and executable in the edge computing infrastructure.
4, network function virtualization
At the heart of network function virtualization is a typical edge computing application because edge computing provides the underlying functionality. Telecom operators are trying to convert their service delivery models by running virtual network functions as part of their or their upper edge computing infrastructure. Running virtual network functions in an edge computing infrastructure can indeed deliver the highest performance and lowest cost/complexity.
Edge computing usage scenario
The so-called edge computing is actually the starting point for the connection between things and humans. Its use scenarios are small to personal-related smartphones, routers, game consoles, set-top boxes, mobile towers, Bluetooth headsets, laptops, embedded car intelligence systems, and large-scale manufacturing. Unmanned robotic arm in the factory, unmanned assembly line, data acquisition of ships, aircraft and outdoor large-scale heavy industry machinery and equipment
Therefore, in this edition of the white paper, the following typical application scenarios are summarized: "Box Cloud" for mobile/financial/remote connection, mobile connection, Network-as-a-Service, universal user Resident equipment (uCPE), satellite communication (SATCOM).
Edge computing provides data collection, computing, and storage services for applications that are far from the central office. The basic feature is to bring the infrastructure closer to the users of the terminal, that is, the site is widely distributed and the edge nodes are connected by the WAN network. A wide variety of edge calculations provide edge services for a wide variety of edge applications. In the entire topology, edge services are a range of more flexible computing services from the edge to the central processing core. And the constant verification in more scenarios in the future will help us to evaluate the existing features that have been applied, and to find that the direction of improvement can be improved.
Edge computing challenge
At the end of the white paper, it also faces the problems and challenges of current edge computing. Although there are currently many instances of edge computing being deployed, there is still a need to address the challenges and limitations that are constantly emerging or that have always existed in order to be truly mass-marketed. For example, for a cross-regional IaaS facility system that can be operated and maintained and relies on WAN interconnection, the edge computing resource management system should provide a set of high-level mechanisms to combine output. In other words, the challenge here is how to revise (and expand as needed) the core services of the IaaS layer to handle the specific needs of edge computing as described above, such as network disconnection/bandwidth, limited functionality of computing and storage devices, lack of management Deployment and more.
As one of the current open source mainstream technologies, OpenStack technology will become an indispensable part of the development of edge computing technology, and will certainly occupy an important seat in the future edge computing overall solution, although "edge computing" It is not and should not be limited to OpenStack components and architecture, but there are many reasons to prove that OpenStack is particularly attractive as an edge computing cloud platform."
The Edge Computing Working Group is currently actively encouraging open source communities to participate in exploring these challenges and possibilities. There is still a lot of work to be done to meet and meet the goals and needs of future technologies. We welcome and encourage the entire open source community to participate in this opportunity to define and develop edge computing.
More information about the edge can be found on the OpenStack official website, and the Chinese version of Edge Computing can be downloaded.
Chinese translator list :
Dr. Song Yiliang, Senior Software Cloud Architect, Intel
Huang Shuquan, Technical Director, Kyushu Cloud
Wang Qing, R&D Manager and Director of OpenStack Foundation, Intel
Should be fool, cloud computing software engineer, Intel
Ding Jianfeng, Cloud Computing Software R&D Manager, Intel