What is Ubiquitous Edge Architecture?
Shenzhen 10Gigabit Ethernet Technology Co.,ltd
Empowering Communication, Securing Networks - Your Trusted Partner for a Connected World
Fancy Wang 2909 2022
LaaS collaboration
In the scenario of edge and cloud collaboration, cloud LaaS is an important resource supplement to edge LaaS, and the two are usually aimed at different types of scenarios. Edge LaaS usually targets relatively limited regional services, such as industrial park intelligent systems, campus information systems, etc.
In such scenarios, front-end data and computing tasks usually do not require extensive access, and task delay requirements are strict. Cloud LaaS is more for wide-area, delay-tolerant business types, such as website hosting.
Edge LaaS provides users with virtualized rental services through network edge computing, storage, network and other resources, and provides service configuration, monitoring and maintenance.
At the same time, considering the complexity of the communication and connection methods of terminal devices in various industries, edge LaaS needs to provide rich interface/protocol capabilities so that it can be applied to a wide range of industry markets. In the collaborative process, the cloud center is responsible for the optimized configuration, scheduling and management strategies of the cloud edge (computing, storage, bandwidth, etc.).
Synergy between PaaS and SaaS
PaaS has quite a few mature products in cloud computing, such as Alibaba Cloud, Microsoft Azure, Amazon AWS, etc. The types of SaaS applications are more diverse, including various online API services, Web services, etc. In the scenario of edge and cloud collaboration, users usually care about the application and data level, and do not care whether their applications and data are on the edge or the cloud.
领英推荐
Similar to LaaS collaboration, the cloud center is often used for non-real-time, long-term big data analysis, and can play an advantage in describing the general characteristics of data, long-term application maintenance and data support. Edge PaaS mainly focuses on short-cycle, real-time data analysis applications, which usually require resource allocation based on scene characteristics.
Distributed D22/D4D Architecture (Ubiquitous Edge Architecture)
D2D usually refers to a new type of communication method. In this method, the communication between devices no longer passes through the backbone network, but forms an ad hoc network through direct device connection and multi-hop to complete the information transmission between devices. The so-called D2D or D4D edge computing means that edge devices become edge computing devices of other devices through D2D. Each device can offload computing to other devices, and can also accept computing requests from other devices.
D2D edge computing is usually suitable for scenarios where infrastructure is lacking or the deployment of infrastructure is very expensive, such as car networking scenarios, battlefield combat scenarios, etc. On the one hand, the lack of edge infrastructure leads to relatively limited resources of the entire D2D edge computing network. On the other hand, each device may generate a very large amount of computing tasks, which is difficult to be completed by a single device.
The D2D edge computing architecture is a purely distributed architecture, and the boundaries between edge and end become very blurred, and each device can be regarded as either an end device or an edge device. This makes the access, routing, addressing, coordination, and distribution technologies in the traditional architecture cannot be directly applied, and there are many system challenges.
For example, in the scenario of intelligent single-soldier combat, each soldier can be regarded as an edge node, and the computing tasks generated by a soldier in an emergency need to be performed collaboratively by multiple nearby soldiers, such as real-time target recognition against the environment Wait.
If a nearby device is selected for collaborative computing according to the traditional task offloading and processing methods, the resource limitation is very likely to cause the task computing to fail to meet the target requirements, thus delaying the military intelligence fighter.
Although the D2D architecture is relatively simple, due to the lack of global information, the lack of stable communication links and physical topology between nodes, the key technologies of D2D edge computing (such as computing offloading, service management, etc.) need to be specified according to the scene. In addition, compared with the traditional architecture, the D2D edge computing architecture also needs to solve problems such as neighbor service discovery and service pricing.