“With the rapid development of the mobile Internet and the Internet of Things, 5G services have shown the characteristics of diverse needs. 3GPP has defined three major scenarios for 5G applications: eMBB (enhanced mobile broadband), URLLC (very reliable low-latency communications) and mMTC (massive machine communications). The eMBB scenario provides high-traffic mobile broadband services with a peak rate of over 10Gbps and bandwidth requirements. Up to tens of Gbit/s, such as high-speed download, high-definition video, VR/AR, etc., meet people’s higher requirements for digital life, which will put tremendous pressure on wireless backhaul networks.
“
With the rapid development of the mobile Internet and the Internet of Things, 5G services have shown the characteristics of diverse needs. 3GPP has defined three major scenarios for 5G applications: eMBB (enhanced mobile broadband), URLLC (very reliable low-latency communications) and mMTC (massive machine communications). The eMBB scenario provides high-traffic mobile broadband services with a peak rate of over 10Gbps and bandwidth requirements. Up to tens of Gbit/s, such as high-speed download, high-definition video, VR/AR, etc., meet people’s higher requirements for digital life, which will put tremendous pressure on wireless backhaul networks. Therefore, it is necessary to sink the business as far as possible to the edge of the network to realize the local distribution of the business. The URLLC scenario provides ultra-high reliability and ultra-low latency communications, such as autonomous driving, industrial control, telemedicine, etc., requiring end-to-end 99.999% high reliability and end-to-end ultra-low latency of less than 1ms, meeting the needs of people’s digital industry. Require. This also requires sinking services to the edge of the network to reduce network delays caused by network transmission and multi-level service forwarding.
In view of the application characteristics of these two scenarios of 5G, the traditional large-scale centralized telecom cloud cannot be satisfied, and the deployment form of the telecom cloud needs to be changed to meet the business needs of different scenarios in the 5G era. The MEC (Mobile/Multi-access Edge Computing) edge cloud can meet this demand. As a key technology for 5G evolution, MEC can be closer to the edge of the customer’s mobile network and provide cloud computing capabilities and IT services for edge applications. It has ultra-low latency, ultra-large bandwidth, localization, high real-time analysis and processing, etc. Features. On the one hand, MEC is deployed at the edge of the network, and the edge service runs on terminal equipment, which provides faster feedback and solves the problem of delay; on the other hand, MEC sinks content and computing power, provides intelligent traffic scheduling, and integrates services. Local distribution and local content caching enable some regional services to be terminated locally, which improves the user’s business experience and reduces the resource occupation of the backbone transmission network and the upper core network. Therefore, in the 5G era, MEC will be the best choice for 5G network edge cloud deployment.
The deployment location of MEC edge cloud
MEC edge cloud does not limit the deployment method of the network. Generally speaking, it can be deployed flexibly according to different business scenarios and delay requirements. It can usually be deployed in the access computer room, the convergence computer room, the core computer room of the city, etc., as follows As shown in the figure:
Figure 1 Deployment location of MEC edge cloud
By deploying the MEC platform at the edge, it is possible to effectively extend the computing power of the cloud from the center to the edge, realize rapid service processing and nearby forwarding, to meet the diverse application scenarios of 5G.
The deployment form of MEC edge cloud
Due to the limitations of the environment (such as space, heat dissipation, load-bearing, etc.) and deployment costs of the computer room at the restricted edge site, the hardware infrastructure of the MEC edge cloud usually uses general-purpose servers based on X86 processors. It features low power consumption and high computing density, making it suitable for deployment in edge computer rooms with lower environmental requirements.
The MEC edge cloud is usually based on a specific cloud platform (such as openstack) to provide a unified virtualized software environment and resource management for upper-layer MEC applications. MEC applications are deployed on edge cloud platforms in the form of virtual machines, and can be deployed flexibly according to different scenarios. For example, edge DCs usually deploy applications that are strongly related to traffic forwarding, including vCPE, OLT-U, DU, BNG -U, UPF/GW-U, MEP, etc., to meet the business needs of large bandwidth and low latency.
ZTE MEC edge cloud deployment plan
In order to better meet the needs of multiple business scenarios in the 5G era, ZTE has proposed a complete 5G-oriented MEC edge cloud deployment plan. The plan has the following 5 highlights:
1. Deploy a lightweight edge cloud platform
Due to the limited environment of the edge site computer room, the deployment scale of MEC edge cloud is usually small, the number of servers deployed at a single site is small, the hardware resources that can be provided are limited, and the installation of infrastructure platforms such as Openstack and management modules will take up a lot Resources, resulting in waste of resources on the edge cloud. Therefore, the MEC edge cloud needs to be “thinned”, and lightweight deployment is required to reduce the resources occupied by the platform and management part, and improve the resource utilization of the edge cloud. “Slimming” mainly includes the following measures:
• First, we must tailor Openstack to keep only necessary components and remove redundant components to reduce the occupation of computing and storage resources on the edge cloud.
• Secondly, the dual-core solution of virtual machine + container can be used to achieve lightweight deployment of upper-layer applications, which is convenient for rapid deployment and upgrade.
• In addition, the thinned Openstack is co-located with the computing nodes on the edge cloud, and does not monopolize physical resources, so as to reduce the resource occupation of the control part.
ZTE’s lightweight MEC edge cloud deployment plan is shown in Figure 2. By deploying lightweight Openstack, the number of components is reduced by 60%, the number of resources required for configuration is reduced by 75%, and the number of management resources is reduced by 10%. This greatly reduces the resource requirements for the edge cloud, and greatly improves the management efficiency of the MEC edge cloud.
Figure 2 Lightweight MEC edge cloud deployment plan
In addition, in order to improve the adaptability and resource density of MEC edge cloud deployment, ZTE and Intel jointly launched a MEC edge cloud dedicated server platform OEP600 (Open Edge Platform 600). The platform uses a small all-in-one design with a chassis depth of only 450mm; the CPU uses Intel’s latest Xeon scalable processor to provide stronger computing performance for edge computing; at the same time, it supports wide temperature work, strong heat dissipation, easy maintenance, and satisfies The various environmental requirements for MEC deployment achieve the best match between performance and cost. At the 2019 MEC Technology and Industry Development Summit, ZTE’s MEC edge cloud server platform OEP600 won the “2018-2019 MEC Technology Innovation Award”, fully demonstrating ZTE’s leading innovation capabilities and technological level in the field of edge computing.
2. Deploy a unified MEC edge cloud management system
Because the scale of the edge cloud is usually small, the number is large, and the location is scattered, which brings great complexity to the planning, deployment, operation and maintenance, and operation of the edge cloud. Therefore, it is necessary to deploy the unified management of the MEC edge cloud at the upper-level convergence site. The platform (shown in Figure 2) provides unified management of lower-level edge sites, and only compute nodes and storage nodes are deployed on each edge cloud, reducing the resource occupation of management modules.
The unified management of MEC edge cloud is mainly manifested in two aspects: resource management and operation and maintenance management:
• Unified resource management
Unified management and allocation of resource pools (such as computing, storage, network, etc.) on all edge nodes, and provide a unified interface to centrally monitor the topology, alarm, performance, capacity and other information of physical resources on each edge node, and provide infrastructure The administrator provides fault location methods such as logs and alarm analysis. NFVO is only deployed on the upper-level aggregation nodes, directly docking with the edge cloud unified management platform, avoiding docking with the resource pools of all edge nodes, and uniformly orchestrating and deploying the virtual machines (VM)/containers deployed on all edge nodes.
• Unified operation and maintenance management
Provide a unified operation and maintenance management platform for VIM in each edge cloud, including site management, user/tenant management, function configuration, image distribution, centralized backup, upgrade/patch management, inspection, API distribution, etc.; provide unified FCAPS management , Unified alarm, configuration and performance statistics; provide intelligent and simple automation tools to quickly implement installation and upgrades, rapid inspections and deployment checks, rapid fault analysis and location, log analysis, etc., to improve operation and maintenance efficiency.
3. Deploy the 5G user plane UPF sinking to achieve local diversion
In order to meet the large bandwidth and low latency characteristics of 5G application scenarios, MEC edge cloud will be deeply integrated with the 5G network architecture during deployment. Its business offloading, policy control, Qos guarantee and other functions will all be implemented through standard 5G network functions. . Based on the C/U separation architecture of the 5G core network, the user plane function UPF (User Plane Function) needs to be deployed at the edge of the network to reduce transmission delay and achieve local diversion of data traffic. Control plane functional network elements such as SMF are centrally deployed in the central DC, which is convenient for unified control of UPF deployed in MEC, unified configuration and distribution strategy.
The 5G user plane UFP sinking deployment to achieve local offloading scheme is shown in Figure 3. The distribution rules required by the UPF on the local MEC are notified to the PCF through the interface, and the PCF configures the distribution strategy to the SMF. The SMF performs centralized scheduling of all traffic. The UL CL “uplink classification” or IPv6 multi-homing scheme can be used to realize the edge UPF selection. The local traffic that needs to be split is offloaded through the local edge UPF, and the non-local traffic is sent to the central UPF through the local UPF for processing, so as to avoid all traffic bypassing the central network, reducing the pressure of backbone network transmission and network construction costs, and increasing the network Carrying efficiency of packet data and user service experience.
Figure 3 5G UPF local offloading solution
4. Provide hardware acceleration for MEC edge cloud to improve performance
MEC edge cloud hardware generally uses x86 general-purpose servers, and x86 general-purpose servers have low processing performance for specific business requirements, resulting in low cost performance and unable to meet the commercial deployment requirements of 5G scenarios. Different software and hardware acceleration solutions need to be considered for different services. :
• For computationally intensive services: such as 5G CU PDCP air interface encryption and decryption processing, MEC positioning algorithm, which consumes a lot of CPU and requires dedicated hardware acceleration
• For traffic forwarding services: such as 5G UPF/GW-U, MEC local offloading, CDN, BRAS-U and other services have high requirements on network forwarding capabilities, and need to accelerate data forwarding software and hardware
• For video-related services: services such as AR/VR, video live broadcast and other services require hardware acceleration for video rendering and transcoding;
• AI field: the training and inference operations involved need to introduce GPU for hardware acceleration
ZTE’s hardware acceleration solution currently supports FPGA-based GTP business acceleration, GPU-based video and audio business acceleration, and QAT encryption and decryption acceleration. Especially for the 5G user plane UPF, smart network card hardware acceleration is used to offload most of the traffic from the CPU to the smart network card to obtain higher forwarding performance. Especially at the MWC World Mobile Conference Shanghai exhibition held in Shanghai, ZTE successfully demonstrated a 5G user plane UPF solution based on an open edge hardware acceleration platform, which reduces the forwarding delay by 90% compared with a virtualized UPF without acceleration. , The throughput is increased by 200%, and the power consumption can be reduced by 55%, which better meets the special requirements of 5G URLLC and eMBB for the forwarding capability of edge data centers.
5. Provide API to realize 5G network service capability opening
The openness of business capabilities is another major feature of MEC applications. MEC is deployed at the edge of the network to provide convenient conditions for real-time perception and acquisition of wireless network information, and then open this information to third-party business applications through open interfaces, which can optimize business applications, enhance user experience, and achieve deep network and business integration.
In order to realize the opening of 5G network capabilities, edge computing applications (ME APP) are introduced into the edge computing system, which interact with the 5G network in real time through NEF (Network Exposure Function). On the one hand, NEF passes the UE and service flow related measurement information, such as UE real-time location, radio link quality, roaming status, etc., to the MEC server. Based on the above measurement information, the MEC server uses intelligent analysis and abstraction to analyze the service performance of the application. Optimize (for example, adjust the video playback bit rate) to improve the quality of service; on the other hand, NEF passes the perceived application service-related information, such as business duration, business cycle, and mobile mode, to the network, which is provided by network-aware analysis applications. Information to further optimize its UE resource configuration (for example, assign appropriate bandwidth resources to VIP users) and session management.
The open architecture of ZTE’s MEC capabilities is shown in the figure below. Relying on the MEC edge service platform, open network capabilities (LBS location capabilities, RNIS wireless network information capabilities, QoS capabilities, bandwidth capabilities, etc.) to provide users with a differentiated experience and tap network value. , Increase the value-added income of operators. At the same time, through the research and standardization of open interfaces, the development and launch of innovative business applications will be accelerated to create a good MEC industry ecological chain.
Figure 4 MEC capability opening architecture
ZTE’s 5G-oriented MEC solution combines platform design, virtualization, MEP, hardware acceleration and other software and hardware technologies with 5G network architecture to provide a lightweight, unified management, high-performance, flexible and open MEC Edge cloud enables localized, short-distance, and distributed deployment of applications, services, and content, which to a certain extent solves the business needs of 5G network eMBB, URLLC, mMTC and other technical scenarios, and enhances user experience. ZTE hopes to work with more industry partners to discuss the cooperation model of edge cloud, build a 5G-oriented edge ecosystem, comprehensively promote the commercialization of MEC edge cloud, and jointly promote the vigorous development of 5G edge services.
The Links: 7MBR25SA120B LB064V02-A3
Related posts
