IP Network Slicing Author : Jian Wang, Zhibo Hu, Jie Dong Copyright Author: Jian Wang, Zhibo Hu, Jie Dong Key Contributors: Wei Shao, Ruiqiang Lu Release Date: 2021-09-15 Issue: 01 Copyright © Huawei Technologies Co., Ltd. 2021. All rights reserved. No part of this document may be reproduced or transmitted in any form or by any means without prior written consent of Huawei Technologies Co., Ltd. Trademarks and Permissions and other Huawei trademarks are trademarks of Huawei Technologies Co., Ltd. All other trademarks and trade names mentioned in this document are the property of their respective holders. Notice The purchased products, services and features are stipulated by the contract made between Huawei and the customer. All or part of the products, services and features described in this document may not be within the purchase scope or the usage scope. Unless otherwise specified in the contract, all statements, information, and recommendations in this document are provided "AS IS" without warranties, guarantees or representations of any kind, either express or implied. The information in this document is subject to change without notice. Every effort has been made in the preparation of this document to ensure accuracy of the contents, but all statements, information, and recommendations in this document do not constitute a warranty of any kind, express or implied. Preface Author Introduction Jian Wang: As a senior datacom documentation engineer in Huawei, he has long been engaged in developing technical documentation for key features of datacom products. He has also been responsible for reviewing and delivering guidance documents for key products and once worked as the editor-in-chief of VNF Product Knowledge Overview and Special Topic - IP FPM. Zhibo Hu: Senior technical expert of Huawei Data Communication Protocol Design Dept. He joined Huawei in 2007 and has taken the lead in Segment Routing over IPv6 (SRv6) research and standardization, involving SRv6, Generalized SRv6 (G-SRv6), slice ID-based network slicing solution, and more. He has published more than 50 individual patents and more than 10 drafts in the Internet Engineering Task Force (IETF). He is the deputy editor-in-chief of SRv6 Network Programming: Ushering in a New Era of IP Networks. Jie Dong: Principal engineer and standards representative in Huawei Data Communication Standard & Patent Dept. He joined Huawei in 2007 and has taken the lead in the research and standardization of routing protocols, Virtual Private Network (VPN), SRv6, 5G transport, and other fields. He has published 12 IETF RFC documents and co-authored SRv6 Network Programming: Ushering in a New Era of IP Networks. i Preface About This Book This book describes the background of IP network slicing, explores its technical value and solution, and presents successful deployment cases. This book aims to help you understand the value and technical architecture of IP network slicing. Intended Audience This book is intended for network planning engineers, network design engineers, mid- and senior-level managers at service providers and enterprises, and readers who want to understand cutting-edge IP network technologies. Because network slicing involves many network concepts, readers of this book should be familiar with IP network basics, such as the IP network architecture, Flexible Algorithm (Flex-Algo), and resource reservation technologies. ii Preface Acknowledgments In writing and publishing this book, we received extensive help and support from both inside and outside Huawei. We sincerely thank Jinzhu Chen, Meng Zuo, Zhenbin Li, Zhiqiang Du, Zhaokun Ding, Dawei Fan, Chenxi Wang, Wenjun Meng, Tao Han, Hongkun Li, Fenghua Zhao, Yue Liu, and other leaders and experts from Huawei Data Communication Product Line for their guidance and support. Our thanks also go to Hui Tian, Shujun Han, Danni Ma, and other experts from China Academy of Information and Communications Technology, who not only provided valuable technical guidance but also carefully reviewed the book. This book focuses on the most cutting-edge IPv6 technologies, which are still evolving and deepening. While we have made significant efforts to ensure accuracy, there might be omissions or deficiencies in the book. Your comments and feedback are warmly welcomed. i Preface Table of Contents Chapter 1 Overview of Network Slicing ......................................................................... 1 Chapter 2 Background of Network Slicing .................................................................... 3 2.1 Emerging Diversified Services....................................................................... 3 2.2 Challenges Facing IP Networks.................................................................... 7 2.3 Background of Network Slicing ................................................................. 11 Chapter 3 Benefits of Network Slicing ........................................................................ 13 3.1 Resource and Security Isolation ................................................................. 13 3.2 Differentiated SLA Assurance ..................................................................... 15 3.3 Extremely High Reliability Assurance ...................................................... 16 3.4 Flexible Topology Connection Customization ....................................... 17 3.5 Automated Slice Management .................................................................. 19 Chapter 4 Architecture of Network Slicing ................................................................ 21 Chapter 5 Network Slicing Solutions ........................................................................... 26 5.1 Overview of Network Slicing Solutions ................................................... 26 ii Table of Contents 5.2 Affinity-based Network Slicing Solution ................................................. 28 5.3 Slice ID-based Network Slicing Solution ................................................. 32 5.4 Comparison Between Network Slicing Solutions ................................. 39 Chapter 6 Suggestions on Network Slicing Deployment ....................................... 41 6.1 Deploying Network Slicing Based on Networking Scenarios ........... 41 6.2 Resource Reservation Based on Service Requirements ...................... 46 Chapter 7 Successful Applications of Network Slicing ........................................... 51 7.1 Smart Healthcare — Slice-based Healthcare Private Network ..... 51 7.2 Smart Policing — Slice-based Public Security Private Network .... 54 7.3 Smart Port — Slice-based Port Private Network ................................ 57 7.4 Smart Grid — Slice-based Power Grid Private Network .................. 60 Chapter 8 Technical Prospects of Network Slicing .................................................. 64 iii Table of Contents Chapter 1 Overview of Network Slicing Communication networks are often compared to transport systems, in which data packets are "vehicles" and networks are "roads." These roads become more and more congested as the number of vehicles increases. To alleviate congestion, transportation departments plan lanes and manage traffic based on vehicle types and operation modes. For example, they usually set dedicated lanes for Bus Rapid Transit (BRT) and non-motorized vehicles. This is also the case with a network. The number of connections and the amount of data traffic will increase rapidly in the evolution from connecting everyone to connecting everything. Without intervention, a network will become more and more congested and complex, eventually affecting the service performance of the network. Similar to the transport system, communication networks also require "lane" division and traffic management, both of which can be achieved using network slicing. Network slicing provides multiple logical networks (slices) on the same shared network infrastructure. Each slice serves a specific service type or industry user, as shown in Figure 1-1, and can flexibly define its logical topology, Service Level Agreement (SLA) requirements, reliability, and security level to meet differentiated requirements of different services, industries, or users. 1 Overview of Network Slicing Figure 1-1 Network slicing Carriers can use network slicing to meet differentiated network connection and service quality requirements of different service types or industry users. This not only eliminates the cost of constructing multiple private networks, but also provides highly flexible network services that can be scheduled and allocated on demand based on service requirements, thereby improving carriers' network value and monetization capability and facilitating the digital transformation of various industries. In a broad sense, network slicing is a complete set of solutions, involving the radio access network, IP network, and mobile core network. This book mainly describes network slicing on the IP network. 2 Overview of Network Slicing Chapter 2 Background of Network Slicing Abstract Network slicing is designed to meet emerging differentiated service requirements in the 5G and cloud era. Within only a few years of being proposed, network slicing has been successfully applied in various industries. This chapter analyzes the challenges facing IP networks from the perspective of diversified services in the 5G and cloud era, and describes the background of network slicing. 2.1 Emerging Diversified Services In terms of network connections, 5G changes their attributes while the cloud changes their scope. The development of new 5G services poses more requirements on network connections, such as stricter SLA assurance and ultralow latency. In addition, the development of various cloud services has brought greater flexibility to service access locations. And some cloud services (such as telco cloud) further break the boundary between physical and virtual network 3 Background of Network Slicing devices, integrating services and transport networks. Such changes have reshaped the scope of network connections. With the emergence of diversified services in the 5G and cloud era, different users pose various service quality requirements on networks. 5G In the 5G era, the characteristics of mobile data, massive device connections, and various vertical industry services vary significantly. Services such as mobile communication, environment monitoring, smart home, smart agriculture, and smart metering require huge numbers of device connections and frequent transmission of many small packets. Other services such as live streaming, video uploading, and mobile healthcare require higher transmission rates, while Internet of Vehicles (IoV), smart grid, and industrial control services require millisecond-level latency and near-100% reliability. As such, 5G networks must provide capabilities such as massive access, ultra-low latency, and ultra-high reliability to meet diversified service requirements of users and vertical industries. Based on the main scenarios and service requirements of mobile Internet and the Internet of Things (IoT), the International Telecommunication Union (ITU) has defined three typical 5G application scenarios, as shown in Figure 2-1. 4 Background of Network Slicing Figure 2-1 Typical application scenarios in the 5G era Enhanced Mobile Broadband (eMBB): focuses on bandwidth-intensive services, such as High Definition (HD) video and Augmented Reality (AR). Ultra-Reliable Low-Latency Communication (URLLC): focuses on services that are extremely sensitive to latency and reliability, such as autonomous vehicle and industrial automation services. Massive Machine-Type Communications (mMTC): covers scenarios with high connection density, such as smart city. These scenarios have different kinds of network feature and performance requirements, which cannot be met using a single network. 5 Background of Network Slicing Cloud Services With the rapid development of the cloud and Internet, more and more enterprises are adopting digital transformation, as shown in Figure 2-2. Figure 2-2 Digital transformation of various industries (cloud migration) According to research by the International Data Corporation (IDC), 100% of enterprises will use cloud services by 2025, and 85% of enterprise applications will be deployed on clouds. Through digital transformation, enterprises hope to achieve asset-light operations by gradually migrating their internal IT support systems and production systems to clouds. This will bring greater efficiency and agility to enterprises using cloud services. Cloudification of enterprise applications transforms the deployment of Information and Communications 6 Background of Network Slicing Technology (ICT) for enterprises, reconstructs private line networks between enterprise sites and clouds, between enterprise sites, and between clouds, and reshapes carriers' Business to Business (B2B) services. One-stop cloud-network services are the most critical requirements of enterprise ICT departments. In the huge enterprise ICT market, more and more industry players are creating different solutions to meet user requirements. Public cloud providers tap into the cloud backbone network field to provide one-stop cloud-network services, gradually eliminating the need for site-to-Internet and site-to-site private lines. In addition, vendors involved with Software-Defined Networking in a Wide Area Network (SD-WAN) provide flexible and cost-effective solutions to meet customers' interconnection requirements. These products and services not only transform private line connections, but also provide flexible connection, fast provisioning, and dynamic adjustment capabilities. Consequently, the market share of carriers' traditional private line services is under threat. To maintain competitiveness in the B2B market, carriers must leverage their advantages in networks and provide flexible, agile, and SLA-guaranteed private line services with wide coverage and cloud-network convergence capabilities. 2.2 Challenges Facing IP Networks With the emergence of diversified services in the 5G and cloud era, meeting diversified, differentiated, and complex requirements of various services on an IP network is a new challenge. Ultra-Low Latency IP networks typically consist of access, aggregation, and backbone layers. Because users are unlikely to all use the maximum bandwidth at the same time, the planned bandwidth is converged from the access layer to the aggregation layer (typically at a ratio of 4:1, although this varies depending on carriers) and then to the backbone layer. Through such convergence, the statistical multiplexing capability of IP networks can be fully utilized, allowing resources to be shared and greatly reducing the network construction cost. The downside of this convergence is that a network may encounter high-speed and multiinterface access but low-speed and single-interface output, resulting in congestion. Although routers use large interface buffers to solve the packet loss 7 Background of Network Slicing problem caused by congestion, buffering packets leads to higher queuing latency if congestion occurs. With the emergence of diversified 5G services, such as those shown in Figure 2-3, different services have different requirements on bandwidth and latency. For example, live video services require high bandwidth, and burst traffic is likely to cause instantaneous congestion; and services such as telemedicine, gaming, and precision manufacturing require ultra-low latency. Requirements on strict latency can be met if channels with differentiated latency are provided based on services. Figure 2-3 Diversified new 5G services Security Isolation Some enterprises in vertical industries, such as government, finance, and healthcare, have specific requirements (shown in Figure 2-4) on security and stability of their core services, such as production, manufacturing, and interactive services. To ensure that these core services are not affected by other services (for example, information management and public network services), enterprises usually isolate them by using private networks. However, factors such as construction costs, Operations and Maintenance (O&M), and rapid service expansion have given rise to enterprises seeking new ways to carry their core services while meeting security isolation requirements. With statistical multiplexing in traditional IP networks, services may preempt each other's 8 Background of Network Slicing resources. As such, only best-effort services can be provided, and the security isolation capability cannot be provided. In addition, traditional Multi-Service Transfer Platform (MSTP) private lines are becoming obsolete, but some services they carry, such as financial and government private line services, require security isolation and exclusive resources. Figure 2-4 Key network indicators required by some industry enterprises Extremely High Reliability High-value services require IP networks to provide high availability, which is one of the key network indicators shown in Figure 2-4. Premium enterprise private line services, such as those in government, finance, and healthcare sectors, usually require an availability of up to 99.99%. This increases to 99.999% for 5G services (especially URLLC services). And for mission-critical services related to social and human safety, such as remote control and high-voltage power supply, an extremely high availability of 99.9999% is essential. As such, it is crucial to provide highly reliable private lines on an IP network to carry these services. 9 Background of Network Slicing Flexible Connections With the continuous development of services in the 5G and cloud era, the singular service type evolves toward diverse types, and the singular traffic pattern evolves toward multi-direction patterns. This results in more flexible, complex, and dynamic network connections, as shown in Figure 2-5. As 5G core Network Elements (NEs) are cloudified, User Plane Functions (UPFs) are moved closer to users, and Mobile Edge Computing (MEC) is widely applied, the connections between base stations, between base stations and different network layers of Data Centers (DCs), and between different network layers of DCs become increasingly complex and change dynamically. This requires networks to provide any connection on demand. In addition, because different industries, services, and users have different service scopes and access locations in the network and cloud, customized network topologies and connections are required. Figure 2-5 Complex IP network connections caused by diversified services Refined and Intelligent Service Management With the emergence of diversified services in the 5G and cloud era, various services not only pose differentiated service requirements on networks, but also pose requirements on network services in terms of being dynamic, real-time, and more. On traditional IP networks, service planning is relatively static and the statistics collection and monitoring of network utilization is performed at a granularity of minutes, ignoring traffic burst characteristics at the micro level. 10 Background of Network Slicing This cannot prevent services from affecting each other, guarantee SLAs, or meet requirements of services on dynamic deployment and flexible adjustment. Tenant-level refined and intelligent management are required for services; however, one traditional IP network cannot meet such requirements. 2.3 Background of Network Slicing A traditional shared IP network cannot efficiently provide guaranteed SLAs for all services, let alone network isolation and independent operation. To meet the differentiated requirements of various services on the same network, network slicing is introduced. With network slicing, carriers can build multiple dedicated, virtualized, and isolated logical networks on a general physical network to meet the differentiated requirements of different customers for network connections, resources, and other functions. Figure 2-6 shows an example of network slicing. 11 Background of Network Slicing Figure 2-6 Example of network slicing Network slicing is a new service mode introduced to carrier networks in the 5G and cloud era. With network slicing, a carrier can provide different network slice services for multiple tenants over a shared network infrastructure to meet differentiated network requirements of different industries. And vertical industry customers can use the network as slice tenants. 12 Background of Network Slicing Chapter 3 Benefits of Network Slicing Abstract Based on network slicing, carriers can provide resource isolation, differentiated SLAs, high reliability, flexible topology customization, and automatic slice management to build intelligent cloud networks, helping enterprises achieve digital transformation. 3.1 Resource and Security Isolation Traffic of different industries, services, and users is carried on the same network through different network slices, which need to provide different types and degrees of isolation based on services and customer requirements. In terms of service quality, the purpose of network slice isolation is to prevent a service burst or abnormal traffic in a slice from affecting other slices in the same network, thereby ensuring that services in different network slices do not affect each other. This is especially important for vertical industries, such as smart grid, smart healthcare, and smart port, which have strict requirements on latency and jitter and whose performance is highly sensitive to impacts from other services. In terms of security, information about private line services (such as financial and 13 Benefits of Network Slicing government services) or users in a network slice should not be accessible to users in other network slices. In this case, effective security isolation measures need to be taken between different slices. Based on the degree of isolation, network slices on IP networks provide three levels of isolation: service, resource, and O&M isolation. Service isolation: Different network slices are established for different services in the public network to isolate service connections and access. Note that service isolation cannot guarantee SLAs; rather, it provides isolation for some traditional services that do not have strict SLAs. This means that one network slice may still be affected by another, even if service isolation is used. Resource isolation: Network resources are defined for exclusive use on a perslice basis, or for sharing among multiple slices. Resource isolation is paramount for 5G URLLC services, which usually have strict SLAs and do not tolerate interference from other services. Based on the degree of isolation, resource isolation includes hard isolation and soft isolation. − Hard isolation ensures that slices are provided with exclusive network resources, preventing any interference between services in different slices. For example, FlexE interfaces or channelized sub-interfaces can be used to provide hard isolation for network slices. − Soft isolation allows each slice to use not only a set of dedicated resources, but also some resources shared with other network slices. This allows services to be isolated to some extent without sacrificing certain statistical multiplexing capabilities. For example, Quality of Service (QoS)/Hierarchical Quality of Service (HQoS) can be used to provide soft isolation for network slices. Based on both hard and soft isolation, carriers can select the optimal combination of resource isolation mechanisms between network slices to meet their resource requirements. This allows a single physical network to meet differentiated SLAs. O&M isolation: In addition to service isolation and resource isolation, some tenants require independent O&M of network slices allocated by carriers, similar to using private networks. Network slicing provides O&M isolation through open interfaces on the management plane. 14 Benefits of Network Slicing Take a smart grid scenario as an example. As shown in Figure 3-1, smart grid services are classified as control or collection services. The two types of services have different SLA requirements, and service isolation needs to be provided. Network slicing provides resource and security isolation between smart grid and public network services, as well as isolation between smart grid control and collection services. Figure 3-1 Isolation between different power grid services 3.2 Differentiated SLA Assurance In addition to bringing a sharp increase in network traffic, the rapid development of network services also gives rise to extreme requirements on network performance. Because different industries, services, and users have different SLA requirements on network bandwidth, latency, and jitter, the same network infrastructure needs to meet differentiated SLA requirements in different service scenarios. On a shared network infrastructure, network slicing provides differentiated SLA assurance for different industries, services, and users. Network slicing enables carriers to gradually transform from selling the same services to selling differentiated services for Business to Home (B2H), Business to 15 Benefits of Network Slicing Business (B2B), and Business to Consumer (B2C). As shown in Figure 3-2, a carrier provides differentiated services for tenants in the form of slice offerings. To drive new value growth in the future, carriers will predominantly provide ondemand, customized, and differentiated services. Figure 3-2 Slice as a service, providing differentiated network services for tenants 3.3 Extremely High Reliability Assurance On IP networks, high-value and URLLC services require high availability and millisecond-level failure recovery. SRv6-based network slicing provides local protection technologies, such as Topology-Independent Loop-Free Alternate (TILFA) and midpoint protection, for any failure point on an IP network. These technologies can significantly increase the protection success rate and enhance the reliability of IP network slices. In addition, link failure-triggered switching in each network slice can be controlled within the slice without affecting other slices, as shown in Figure 3-3. 16 Benefits of Network Slicing Figure 3-3 Link failure-triggered switching in a network slice 3.4 Flexible Topology Connection Customization The continuous development of services in the 5G and cloud era gives rise to network connections that are more flexible, complex, and dynamic. As shown in Figure 3-4, network slicing uses Flexible Algorithms (Flex-Algos) to provide ondemand customization of logical network topology connections, meeting differentiated network connection requirements of different industries, services, and users. 17 Benefits of Network Slicing Figure 3-4 Flex-Algo-based flexible customization of topology connections After a logical topology and connection are customized for users in a network slice, the users are aware of only the network slice's logical topology and connection — not the basic network's full network topology. In addition, services deployed in a network slice are limited to the topology corresponding to that network slice. For network slice users, this simplifies the network information that they need to perceive and maintain. And for carriers, this prevents excessive internal information about basic networks from being exposed to network slice users, improving network security. 18 Benefits of Network Slicing 3.5 Automated Slice Management As service types and scales continuously increase, the complexity involved in network management grows rapidly. This means that managing networks manually is no longer a feasible option for carriers. Instead, they need to manage these networks dynamically and efficiently, which requires automated network management technologies. The network slice manager provides fulllifecycle management and tenant-level refined service management for network slices, streamlining the entire process from user intent to service provisioning, as shown in Figure 3-5. Refined slice planning: Slices based on service requirements provide differentiated SLAs, fully meeting the bearer requirements of various customers and preventing wastage of slice resources. Slice automation: Minute-level slice automation based on an intelligent network controller implements fast deployment and on-demand capacity expansion of slices. SLA visualization: An intelligent network controller visualizes the network topology and resources of service-level slices and supports SLA visualization based on In-situ Flow Information Telemetry (IFIT) measurement. 19 Benefits of Network Slicing Figure 3-5 Full-lifecycle management of network slices With the ongoing development of network management automation, intelligence technologies will be widely used in each phase of network slice management to implement intelligent network management. 20 Benefits of Network Slicing Chapter 4 Architecture of Network Slicing As shown in Figure 4-1, the IP network slicing architecture consists of three layers: network slice management layer, network slice instance layer, and network infrastructure layer. 21 Architecture of Network Slicing Figure 4-1 IP network slicing architecture Network Slice Management Layer This layer provides lifecycle management for network slices. To meet the requirements of different services, network slicing divides a physical network into multiple logical network slices. Because this increases the management complexity of network slices, automated and intelligent management of network slices is crucial, involving the planning, deployment, O&M, and optimization of network slices, as shown in Figure 4-2. 22 Architecture of Network Slicing Figure 4-2 Network slice lifecycle management Slice planning: Physical links, forwarding resources, service Virtual Private Networks (VPNs), and tunnels need to be planned for network slices in order to guide network slice configurations and parameter settings. In addition, multiple solutions for network slice planning are provided, such as network-wide slicing based on fixed bandwidth, flexible customization of topology connections, and automatic calculation of slice topologies and required resources based on service models and SLA requirements. Slice deployment: Network slice instances need to be deployed, including creating network slice interfaces and configuring bandwidth, VPNs, and tunnels for network slices. Slice O&M: Functions such as network slice visualization and fault O&M need to be provided. IFIT is used to monitor service latency and packet loss, and telemetry is used to report a network slice's traffic volume, link status, and service quality information in order to show the network slice status in real time. 23 Architecture of Network Slicing Slice optimization: Network slice performance needs to be balanced with network costs to meet SLA requirements through various operations, such as slice forwarding resource prediction and intra-slice traffic optimization. Network Slice Instance Layer This layer enables the instantiation of different logical network slices on a physical network, supports on-demand customized logical topology connections, and associates the logical topologies of network slices with the set of network resources allocated to the slices. In this way, network slices are formed to meet specific service requirements. The network slice instance layer covers VPNs at the overlay layer and Virtual Transport Networks (VTNs) at the underlay layer. VPNs provide logical connections for services within a network slice and can isolate services of different network slices. VTNs, on the other hand, provide logical network topologies for slice service connections, and provide exclusive or partially shared network resources to meet SLAs of network slice services. In this regard, a network slice instance is the integration of a VPN service as the overlay with an appropriate VTN as the underlay. Because there are various overlay VPN technologies that are mature and widely used, the following sections mainly describe the VTN's functions. Network slice mentioned later typically refers to a VTN that carries network slice services. VTN functions can be broken down into data plane functions and control plane functions. Data plane: adds network slice identifiers to data packets so that packets of different network slices can be forwarded according to the forwarding entries of the corresponding network slices. Note that these identifiers are generic and agnostic to various resource partitioning technologies used at the network infrastructure layer. Currently, network slice IDs can be carried using SRv6 Segment Identifiers (SIDs) or dedicated slice IDs in data packets. Control plane: distributes and collects each network slice's attributes (such as the topology and resource) and their status information. In addition, the control plane calculates and provisions routes and paths based on the network slice's topology and resource constraints, mapping service flows of different network slices to corresponding network slice instances on demand. Currently, the network slice topology can be flexibly customized using Flex- 24 Architecture of Network Slicing Algos on the control plane, and path information within network slices can be delivered using SRv6 Policies. Network Infrastructure Layer This layer is a physical network used to create IP network slice instances. To meet services' requirements on resource isolation and SLA assurance, the network infrastructure layer needs to have flexible and fine-grained resource reservation capabilities so that it can partition the physical network's forwarding resources into multiple sets of isolated resources based on a required granularity for allocation to different network slices. Some candidate resource partitioning technologies include Flexible Ethernet (FlexE) sub-interface, channelized subinterface, and Flex-channel. 25 Architecture of Network Slicing Chapter 5 Network Slicing Solutions Abstract This chapter mainly describes the design principles and characteristics of the affinity-based and slice ID-based network slicing solutions, and compares the two solutions. This will help you understand how network slicing implements functions such as resource isolation and differentiated SLA assurance. 5.1 Overview of Network Slicing Solutions Two common network slicing solutions are currently available: affinity-based network slicing solution and slice ID-based network slicing solution. This section briefly describes the two solutions. For further details, see 5.2 and 5.3 . Affinity-based network slicing solution As shown in Figure 5-1, the affinity-based network slicing solution uses an affinity to identify a slice. Each affinity corresponds to one network slice. 26 Network Slicing Solutions Affinities can identify the forwarding resource interfaces of different slices, and each resource interface requires an IP address and SR SID. In the control plane, each slice calculates Segment Routing-Multiprotocol Label Switching (SR-MPLS) and SRv6 Policy paths based on the affinity for service bearing. In the data plane, slice-specific service packets are encapsulated with the SRMPLS label stack or SRv6 Segment Routing Header (SRH) and forwarded hop by hop. This book uses SRv6 as an example to describe the affinitybased network slicing solution. Figure 5-1 Affinity-based network slicing solution Slice ID-based network slicing solution As shown in Figure 5-2, the slice ID-based network slicing solution introduces a globally unique slice ID to identify a network slice. Each slice ID corresponds to one network slice and identifies the forwarding resource interfaces in the slice — there is no need to configure an independent IP address and SR SID for each slice interface. In the control plane, SR-MPLS, SRv6 Best Effort (BE), or Traffic Engineering (TE) Policy paths are calculated based on slice IDs for service bearing. In the data plane, each forwarding node matches the slice resource interface based on the slice ID carried in a 27 Network Slicing Solutions data packet for service forwarding. This book uses SRv6 as an example to describe the slice ID-based network slicing solution. Figure 5-2 Slice ID-based network slicing solution 5.2 Affinity-based Network Slicing Solution Affinity-based network slicing uses the existing control plane and data plane protocol mechanisms to quickly establish and adjust network slices based on service requirements. It is applicable for fast deployment of network slices in legacy networks. 28 Network Slicing Solutions What Is an Affinity? An affinity, also called admin group or color, is a control information attribute of a link. As shown in Figure 5-3, affinities are used to identify links by assigning the links different colors (such as blue and yellow). Links identified with the same color form a logical network topology. Figure 5-3 Using affinities to identify links Affinity-based Network Slicing As shown in Figure 5-4, an affinity is used as a control plane identifier of a network slice, and different affinities are configured on resource reserved interfaces or sub-interfaces corresponding to each network slice. In this way, an independent network slice is planned based on the affinity. 29 Network Slicing Solutions Figure 5-4 Affinity-based network slicing Affinity information (along with other link information) is flooded in the network through the Interior Gateway Protocol (IGP)/Border Gateway Protocol-Link State (BGP-LS) and is reported to the network slice controller. After collecting the link status information of the entire network, the network slice controller may form an independent network slice view based on each affinity, and compute a constrained forwarding path for slice services on a per-slice basis. As shown in Figure 5-5, in the data plane, different SRv6 End.X SIDs need to be allocated to resource interfaces or sub-interfaces reserved for different network slices. In this way, each forwarding node on the network can determine the interface or sub-interface resources for packet forwarding based on the SRv6 SIDs. 30 Network Slicing Solutions Figure 5-5 SRv6-based data plane The network slice controller computes an explicit path based on slice constraints. The path can then be orchestrated into a SID list composed of SRv6 SIDs relating to interfaces or sub-interfaces. This SID list can explicitly indicate the forwarding path of packets and a group of reserved forwarding resources on the path in the SRv6 network. The controller uses a BGP SR Policy to deliver the SRv6 explicit path of each slice to the ingress and steer the services — such as Layer 2 Virtual Private Network (L2VPN) and Layer 3 Virtual Private Network (L3VPN) services — planned in the slice to the SRv6 Policy path of the corresponding slice. As shown in Figure 5-6, if the destination address of a service matches the endpoint of an SRv6 Policy and the service preference (identified by the color extended community attribute of the corresponding VPN route) is the same as that of the SRv6 Policy, the service can be steered to the SRv6 Policy for forwarding. To ensure resource isolation between different slices and provide differentiated paths for different services within a slice, an SRv6 Policy restricts the forwarding of service packets using the paths and reserved resources within a slice. This makes it possible to meet the SLA requirements of different slice users and different services within a slice. 31 Network Slicing Solutions Figure 5-6 Steering services to a specified SRv6 Policy path based on an affinity 5.3 Slice ID-based Network Slicing Solution Slice ID-based network slicing introduces dedicated global slice IDs to data packets in order to identify network slices with a simpler and more straightforward approach. This differs from affinity-based network slicing, which uses SRv6 SIDs to identify network slices. In affinity-based network slicing, each device that reserves resources for network slices must allocate a different SRv6 locator and a set of SIDs to each network slice. Consequently, the number of SRv6 locators and SIDs to be allocated increases rapidly as the number of network slices increases, bringing challenges to network planning and management. This also multiplies the amount of information to be advertised by the control plane and the number of forwarding entries in the data plane, bringing scalability problems to the network. By using dedicated global slice IDs, slice ID-based network slicing prevents the number of SRv6 locators and SIDs 32 Network Slicing Solutions from multiplying with the number of slices, effectively relieving the scalability pressure caused by the increase of network slices in the control and data planes. What Is a Slice ID? The biggest change brought by network slicing is the shift from a traditional network consisting of one physical plane to a three-dimensional network consisting of many logical planes. As shown in Figure 5-7, each network node in a logical plane is identified using a unique IP address, which is essential for packet forwarding. In a multi-plane three-dimensional network, however, this one-dimensional identification method causes major issues. Because different slices have different network topologies or network resources, this method requires a different IP address to be allocated to each node in each slice for identification. For example, if there are 1000 network nodes and 200 network slices need to be created, 200,000 IP addresses need to be planned. This brings major challenges to network deployment and performance. 33 Network Slicing Solutions Figure 5-7 Network slice address identification model (one-dimensional identification) To resolve this problem, two-dimensional addressing is introduced in order to identify network slices of different logical planes. As shown in Figure 5-8, in the two-dimensional identification method, the IP address of a physical network node and a network slice ID are used together to uniquely identify a logical node in the network slice. In this way, only one set of IP address identifiers is required, and address planning and configuration do not need to be separately performed for each network slice — regardless of the number of network slices planned for a network. In addition, using a two-dimensional address identifier can significantly reduce the number of routes in a network slice, easily supporting Klevel network slices. 34 Network Slicing Solutions Figure 5-8 Network slice address identification model (two-dimensional identification) To support a two-dimensional address identifier, a global network slice ID needs to be added to data packets. Typically, the Hop-by-Hop (HBH) Options header of an IPv6 packet carries the global data plane identifier (network slice ID) of a network slice. As shown in Figure 5-9, the slice ID specifies the slice over which the packet is carried. 35 Network Slicing Solutions Figure 5-9 Format of an IPv6 packet encapsulated with an HBH Options header Slice ID-based Network Slicing Traditional IPv6 forwarding is based on destination addresses. Slice ID-based network slicing reuses the addresses of the basic network, without requiring an additional IPv6 address to be separately allocated to each slice. Slice IDs that are globally planned and allocated are used to identify the forwarding resources allocated by network devices to network slices. In this case, the default and service network slices differ only in forwarding resources and data plane identifiers. In the data plane, network devices use the destination address and a slice ID to instruct packet forwarding in a network slice. The destination address is used to address a packet forwarding path, whereas the slice ID is used to select forwarding resources corresponding to a packet. In the control plane, different network slices can reuse protocol sessions and route calculation, reducing the pressure caused by an increase in the slice scale on the control plane. As shown in Figure 5-10, nine network slice instances are created: three each on DeviceA, DeviceB, and DeviceC. An independent slice ID is used to identify the resource interface or sub-interface allocated to each network slice on a physical port. On a network node, all network slices share the same IPv6 address and control-plane protocol session. 36 Network Slicing Solutions Figure 5-10 Slice ID-based data plane For a slice ID-based network slice, a network device needs to generate two forwarding tables. One is a routing table, which is used to determine the Layer 3 outbound interface based on the destination address of a packet. The other is a slice interface's slice ID mapping table, which is used to determine a slice's reserved resources on the Layer 3 interface based on the slice ID in a packet. As shown in Figure 5-11, after a service packet reaches a network device, the network device searches the routing table based on the destination address in order to obtain the next-hop device and Layer 3 outbound interface. The device then searches the slice interface's slice ID mapping table based on the slice ID to determine reserved resources (sub-interfaces or channels) on the Layer 3 outbound interface. Finally, the device uses the corresponding sub-interface or channel to forward the service packet. 37 Network Slicing Solutions Figure 5-11 Steering services into a specified SRv6 Policy path based on a slice ID The slice ID-based network slicing solution has the following advantages: Multiple network slices reuse the same address identifier, simplifying network slice deployment. Multiple network slices share the routing table. Slice IDs are used for the resource mapping table lookup to provide differentiated forwarding of network slices, reduce the scale of routes in slices, and improve convergence performance. Topology and resource decoupling is achieved to reuse the slice topology as much as possible, reduce the overhead caused by the protocol used by the controller to maintain multiple slice topologies, and increase the slice scale. 38 Network Slicing Solutions 5.4 Comparison Between Network Slicing Solutions Table 5-1 compares the affinity-based network slicing solution with the slice IDbased network slicing solution. Table 5-1 Comparison between network slicing solutions Item Affinity-based Slicing Solution Network Slice specifications 16 (maximum) Thousand-level Forwarding plane isolation FlexE/Channelized sub-interface FlexE/Channelized technology Slice ID-based Network Slicing Solution sub- interface/Flex-channel SLA assurance effect Strict assurance Strict assurance Configuration complexity Complex Simple Whether IP addresses and Yes No Pre-deployment Pre-deployment Layer 3 protocols need to be configured for service slice interfaces Service slice deployment mode + on- demand deployment (ondemand slicing) SRv6 working mode Whether a controller is SRv6 Policy SRv6 BE/SRv6 Policy Yes Yes A small number of slices are Strict required, and fast deployment required for thousand-level can be implemented in legacy users, and massive network networks. slices are required. required Application scenario 39 Network Slicing Solutions SLA assurance is Item Affinity-based Slicing Solution Evolution path The affinity-based Network network Slice ID-based Network Slicing Solution N/A slicing solution can evolve to the slice ID-based network slicing solution. Currently, although the affinity-based network slicing solution can be quickly deployed on live networks, its limitations include supporting only a small number of slices and involving complex configurations. The slice ID-based network slicing solution does not have such limitations and is therefore the preferred choice for large-scale network slice deployment. 40 Network Slicing Solutions Chapter 6 Suggestions on Network Slicing Deployment Abstract The deployment of network slicing usually requires the help of a network controller. Before deploying network slicing, you need to consider factors such as networking scenarios and SLA requirements. This chapter provides suggestions on deploying network slicing based on networking scenarios and reserving resources based on SLA requirements. 6.1 Deploying Network Slicing Based on Networking Scenarios Different network slicing solutions have different deployment modes in different networking scenarios. Before deploying network slicing, you need to determine which solutions best meet your networking requirements. 41 Suggestions on Network Slicing Deployment Network Slicing in Different Networking Scenarios According to the network connection model, there are three typical networking scenarios: Multipoint-to-Multipoint (MP2MP), Point-to-Point (P2P) private line, and hybrid network modes. As shown in Figure 6-1, network slices deployed in the three networking scenarios are called MP2MP network slices, P2P private line network slices, and hybrid network slices. For MP2MP network slices, the entire physical network or a portion of it can be sliced, and nodes in a MP2MP network slice are fully meshed. Typical MP2MP network slices include carrier self-operating service slices, industry-specific slices, and VIP customer-specific slices. Typically, network resources cannot be shared between different MP2MP network slices, but can be shared between different connections in the same slice. An MP2MP network slice usually requires multipoint-to-multipoint interconnection, resulting in many connections and complex connection relationships. For example, in an MP2MP network slice with 1000 nodes, to implement interconnection between any two nodes, about 1,000,000 point-to-point explicit paths need to be established. The complexity brought by such a large number of paths puts great pressure on network performance. As such, it is recommended that Flex-Algos be used to customize slice topologies and provide distributed path computation for different MP2MP network slices. A Flex-Algo is a customized algorithm for constrained path computation. With Flex-Algos, you can define algorithm values and a series of parameters (including metric types, algorithm types, and link constraints) to flexibly customize topologies and path computation rules. In this way, network nodes can perform distributed path computation based on constraints, reducing the cost of computing and maintaining a large number of tunnels. For P2P private line network slices, such as government and enterprise as well as enterprise site-to-site private line network slices, slicing is implemented based on specified service access points, and such private line network slices usually require exclusive bandwidth resources. A private line network slice usually requires interconnection only between limited service access points, and the connection relationship between access points is relatively fixed. As such, while a single private line network slice has only a limited number of connections, the number of private line network slices on an entire network is relatively large. If each P2P private line network slice is deployed using a Flex-Algo, a network needs to support a large number of Flex-Algos, placing great pressure on network performance. For this reason, it is recommended that SRv6 Policies be 42 Suggestions on Network Slicing Deployment used to provide explicit paths for connections in P2P private line network slices to implement differentiated forwarding. A hybrid network slice is a combination of an MP2MP network slice and a P2P private line network slice, and as such has both of their characteristics. Specifically, a hybrid network slice combines a Flex-Algo and SRv6 Policy: a FlexAlgo is used to customize the slice topology and provide distributed differentiated path computation, whereas an SRv6 Policy is used to provide deterministic forwarding paths for some service flows in a slice. Figure 6-1 Networking modes of network slices While both the affinity- and slice ID-based network slicing solutions can meet the requirements of the three networking scenarios, their implementation processes differ. The following provides suggestions for deploying the two solutions in the MP2MP network slice and P2P private line network slice scenarios. A hybrid network slice is a combination of an MP2MP network slice and a P2P private line network slice, and details are not described herein. Application of the Affinity-based Network Slicing Solution in Networking Scenarios The affinity-based network slicing solution uses an SRv6 Policy to explicitly specify a service path with guaranteed resources between two endpoints on a network. This solution can meet the deployment requirements of P2P private line network slices and can also be used for MP2MP network slices. An SRv6 Policy is 43 Suggestions on Network Slicing Deployment used between any two network nodes to specify a service path with guaranteed network resources, and then a set of SRv6 Policies between each pair of multiple network nodes forms an MP2MP network slice with guaranteed network resources. When there are a large number of service connections in an MP2MP network slice and the connection relationships dynamically change, a large number of SRv6 Policy paths need to be calculated and delivered for the affinity-based SRv6 Policy network slice. This may put much pressure on the performance of the controller and network devices. As such, this type of MP2MP network slice can use the affinity- and Flex-Algo-based slicing solution, as shown in Figure 6-2. In this solution, Flex-Algos are used to define different network slices' topologies and path computation constraints, which are flooded to each network device through an IGP. Network devices can then compute SRv6 BE forwarding paths that meet slice constraints based on the topologies, and use the reserved network resources that are identified by affinities of the network slice for service packet forwarding. In this case, most service packets in a network slice are forwarded using Flex-Algo-based SRv6 BE paths with guaranteed resources. SRv6 Policies are mainly used to provide explicit paths for some services in the slice, reducing the number of SRv6 Policies required by the slice, so that MP2MP network slice services can be delivered more efficiently. Figure 6-2 Affinity- and Flex-Algo-based MP2MP network slice solution 44 Suggestions on Network Slicing Deployment Application of the Slice ID-based Network Slicing Solution in Networking Scenarios In the slice ID-based network slicing solution, slice IDs are globally planned and allocated to identify the subset of forwarding resources that network devices allocate to network slices. This can meet the requirements of both MP2MP network slices and P2P private line network slices, but the implementation solutions differ between them. An MP2MP network slice consists of a large number of connections and a complex network topology. To cope with this situation, slice IDs and Flex-Algos can be combined to provide topology customization and resource guarantee for network slices, as shown in Figure 6-3. After Flex-Algos are enabled on network nodes, the network nodes perform path computation according to the algorithm parameters defined by the Flex-Algos. Then, based on the link constraints defined in Flex-Algos, a physical network can be divided into different logical topologies to meet differentiated topology customization requirements of network slices. In addition, Flex-Algos allow different metric types to be used to compute differentiated paths for network slices with the same topology, meeting differentiated SLA requirements of network slices. After Flex-Algos determine slice topologies and packet forwarding paths, slice IDs are used to identify the network resources reserved for the slices and used during packet forwarding. Figure 6-3 MP2MP network slices based on slice IDs+Flex-Algos As shown in Figure 6-4, an SRv6 Policy is used to specify a service path with guaranteed resources between two endpoints on the network. End SIDs and End.X SIDs are allocated to the network nodes and Layer 3 interfaces, respectively, while a slice ID identifies a private line tunnel with guaranteed 45 Suggestions on Network Slicing Deployment resources on the network. Multiple SRv6 Policies can use the same End/End.X SID to specify explicit paths, and different slice IDs to identify different resources reserved for the slices on the paths. This ensures that differentiated SLA requirements of different P2P private line network slices can be met. Figure 6-4 Slice ID-based P2P private line network slices 6.2 Resource Reservation Based on Service Requirements Resource reservation technology is key to providing differentiated SLA assurance for network slicing solutions. It partitions forwarding resources in a physical network into multiple mutually isolated resource groups for different network slices to use. This ensures that resources are available to meet service requirements in network slices, and prevents or controls resource contention and preemption between different network slices. This section describes common resource reservation technologies (including FlexE interface, channelized subinterface, and Flex-channel) in network slicing solutions. In actual network deployment, proper resource reservation technologies can be selected based on different service requirements for refined allocation of network resources. FlexE Interface FlexE technology uses FlexE shim to pool physical interface resources based on slots. A high-bandwidth physical interface is flexibly divided into several sub- 46 Suggestions on Network Slicing Deployment channel interfaces (FlexE interfaces) based on a slot resource pool, implementing flexible and refined management of interface resources. A FlexE interface is equivalent to a physical interface, and its bandwidth resources are strictly isolated from those of other FlexE interfaces. FlexE interfaces have extremely little latency interference with each other and can provide ultra-low latency. This makes FlexE interfaces suitable for carrying URLLC services that have strict requirements on latency SLA, such as differential protection services of power grids. FlexE interface-based slice resource reservation has the following characteristics: Guaranteed performance: The latency is stable and no packet loss occurs after slicing; hard isolation is implemented between slices; bandwidth is guaranteed; and services in different slices do not affect each other. Fine-grained slicing: Huawei supports a minimum slicing granularity of 1 Gbit/s with FlexE. In contrast, only 5 Gbit/s slicing granularity is supported in the industry. Scalable slicing: When used with other resource reservation technologies, such as channelized sub-interface or Flex-channel, FlexE supports hierarchical slicing to meet requirements for more complex service isolation. Instant slicing: Slices can be deployed in minutes for fast service deployment. Slice resources can be pre-deployed through an intelligent network controller or deployed on demand for services. Reliable slicing: The slice bandwidth is dynamically adjusted, and services are stable. Intelligent O&M capabilities such as slice-based SLA visualization are supported. Channelized Sub-interface Channelized sub-interfaces are based on the sub-interface model. Leveraging the HQoS mechanism, bandwidth can be flexibly allocated by configuring independent channelized sub-interfaces for network slices. Each network slice is allocated exclusive bandwidth and a dedicated scheduling tree to reserve resources for slice services. A channelized sub-interface is equivalent to an independent "lane" in a road. Each of these lanes is assigned to each network slice on a network device. The lane between different network slices is fixed, and cannot be changed during service traffic transmission. This ensures strict isolation of services in different slices, and effectively prevents resource 47 Suggestions on Network Slicing Deployment preemption between services when traffic bursts occur. In addition, flexible lanes can be assigned within the fixed lane of each network slice, allowing differentiated scheduling to be performed in the same slice based on the priority of packets. Channelized sub-interfaces are independent logical interfaces that reside on a physical interface. They are suitable for creating logical networks and are usually used to provide MP2MP network slice services with guaranteed bandwidth. Channelized sub-interface-based slice resource reservation has the following characteristics: Strict resource isolation: Based on the sub-interface model, resources are reserved in advance to prevent slice services from preempting resources when traffic bursts occur. Fine bandwidth granularity: Channelized sub-interfaces can be used together with FlexE interfaces, dividing a high-rate interface into sub-interfaces with low bandwidth. The sub-interfaces can be used to provide industrial network slices. Flex-channel A Flex-channel provides a flexible and fine-granularity interface resource reservation mode. In contrast to a channelized sub-interface, a Flex-channel does not have a sub-interface model and is easier to configure. As such, a Flexchannel is more suitable for scenarios where network slices are quickly created on demand. Flex-channel-based slice resource reservation has the following characteristics: On-demand slicing: Service-based slicing requirements are quickly delivered by a controller to implement on-demand slicing. Massive network slices: A Flex-channel supports a minimum bandwidth granularity of 1 Mbit/s, meeting the slice bandwidth requirements of enterprise users. 48 Suggestions on Network Slicing Deployment Comparison Between Different Resource Reservation Technologies Table 6-1 compares different resource reservation technologies. Table 6-1 Comparison between resource reservation technologies Item FlexE Interface Channelized interface Isolation Exclusive use of Traffic TM Manager (TM) resources; reservation; port resource isolation resource sharing Latency A A assurance increase of up to 10 µs single-hop latency Sub- resource single-hop port Flex-channel TM port resource sharing latency A single-hop latency increase of up to 100 increase of up to 100 µs µs Granularity 1 Gbit/s 2 Mbit/s 1 Mbit/s Application Industrial network slice Industrial network slice Enterprise and enterprise MP2MP network network enterprise scenario resource reservation; slice deployment) (pre- network P2P slice and MP2MP slice (on- demand slicing) Different resource reservation technologies can be used together, as shown in Figure 6-5. Carriers usually use FlexE interfaces or channelized sub-interfaces to reserve coarse-grained slice resources for specific industries or service types. These are then further divided into Flex-channels to reserve fine-grained slice resources for different enterprise users. 49 Suggestions on Network Slicing Deployment Figure 6-5 Combination of different resource reservation technologies Network slices with hierarchical scheduling are used to provide flexible and refined resource management. For example, on a network with 50 Gbit/s bandwidth in the access ring and 100 Gbit/s bandwidth in the aggregation ring, FlexE interfaces can be used on the access and aggregation rings to reserve 1 Gbit/s and 2 Gbit/s bandwidth, respectively. This implements hard isolation of services and meets the requirements of a vertical industry on isolation and ultralow latency. After entering the aggregation ring from multiple access rings, services in the same slice can share the 2 Gbit/s bandwidth reserved for the slice in the aggregation ring. Different service types or users of the vertical industry can continue to use channelized sub-interface or Flex-channel technology in a FlexE interface of a slice to perform refined resource reservation and scheduling. This maximizes statistical multiplexing of resources while meeting requirements on slice isolation and SLA assurance. 50 Suggestions on Network Slicing Deployment Chapter 7 Successful Applications of Network Slicing Abstract To meet differentiated SLA requirements in the 5G and cloud era, carriers support different services through network slicing, helping enterprises achieve digital transformation. This chapter describes the successful applications of network slicing in four scenarios: smart healthcare, smart policing, smart port, and smart grid. 7.1 Smart Healthcare — Slice-based Healthcare Private Network Requirement Introduction As shown in Figure 7-1, smart healthcare is a healthcare service system that uses network technologies to implement prevention, consultation, diagnosis & 51 Successful Applications of Network Slicing treatment, rehabilitation, health care, and more. During the promotion of smart healthcare, high-speed broadband networks and interconnected private lines covering urban and rural healthcare organizations are crucial for supporting the smart healthcare system. Figure 7-1 Different services of smart healthcare To support cloudification and interconnection for services of hospitals at all levels in urban and rural areas, as well as implement telemedicine anytime and anywhere, the healthcare private network needs to have the following capabilities: Full-mesh connection: City-level tertiary A hospitals serve as centers and establish hub-spoke interconnections with county-level hospitals. This enables service channels to be quickly established between any two healthcare organizations. High bandwidth: The bandwidth of each village clinic and community health station is increased to 300 Mbit/s. The bandwidth of township health centers and community health centers is increased to 500 Mbit/s. The bandwidth of the county-level or higher-level health and family planning commissions and secondary and tertiary hospitals is increased to 1 Gbit/s. 52 Successful Applications of Network Slicing Ultra-low latency: Core healthcare systems are migrated to the cloud. The service latency of systems such as image archiving & communication systems and hospital information management systems is less than 20 ms and remains stable, providing service experience the same as local services. Smart Healthcare Based on Network Slicing The "Healthcare Cloud Network" product provided by the network slice private network can implement one network for multiple purposes, as shown in Figure 7-2. Figure 7-2 Healthcare private network based on network slicing This private network is the first in the industry to deploy slicing and SRv6 technologies across the entire network. It has the following characteristics: Network slices for hard isolation: Resources of different slices are independent of each other. The SLAs of services in a slice can still be guaranteed, even if services in other slices are congested. 53 Successful Applications of Network Slicing Fast provisioning and agile O&M: NCE provides E2E network slice lifecycle management, and slices and SLAs are visualized and controllable. One network for multiple purposes: One healthcare private network can be further sliced to provide multiple service slices, enabling one network for multiple purposes and high Return On Investment (ROI). 7.2 Smart Policing — Slice-based Public Security Private Network Requirement Introduction Public security systems face challenges such as insufficient police resources and low law enforcement efficiency. With the 5G network, public security systems can implement smart policing such as police drone, video surveillance, AR patrol, and comprehensive intelligence command system, as shown in Figure 7-3. Smart policing can implement ground-air multi-dimensional patrol within a jurisdiction, making up for insufficient police resources and improving comprehensive law enforcement efficiency. 54 Successful Applications of Network Slicing Figure 7-3 Different policing services of smart policing To implement smart policing, the public security information network needs to have the following capabilities: High security and reliability: Uploaded policing surveillance data has high security requirements and must be fully isolated from public data transmission channels. High bandwidth: Drone patrol, smart police cars, and high-altitude cameras require 4K HD surveillance, and 4K video recording requires real-time transmission (20 Mbit/s to 40 Mbit/s uplink bandwidth per channel). Guaranteed latency: In the hotspot areas of B2C user traffic, E2E resource reservation can be used to guarantee the latency of public security services. Smart Policing Based on Network Slicing E2E network slicing can be used to provide a private network for policing services to meet the security isolation requirements of policing and public user services. 55 Successful Applications of Network Slicing Figure 7-4 Smart policing based on network slicing The network slicing-based characteristics: smart policing solution has the following High security and reliability: An independent E2E channel for the policing service slice is deployed, where Radio Bearer (RB) or Allocation and Retention Priority (ARP) is deployed on the Radio Access Network (RAN) to reserve resources; FlexE slicing is deployed on the IP network to reserve resources; and a standalone UPF is deployed on the core network). This isolates policing services from public services and ensures the security of uploaded policing surveillance data. High bandwidth: High-bandwidth slices are deployed (10GE for base stations, 50GE for the IP network, and 100GE for the core and aggregation layers) to meet the upstream bandwidth requirements of videos and implement realtime uploading of multi-channel HD videos. Guaranteed latency: FlexE slicing is deployed to implement hard pipe isolation and guarantee the stable latency of policing services. Visualized slice SLA: IFIT is deployed to visualize the SLA of service slices, quickly locating network faults. 56 Successful Applications of Network Slicing 7.3 Smart Port — Slice-based Port Private Network Requirement Introduction The port industry is a heavy machinery industry. The main operation site of a port is located in the outdoor yard, which occupies a large area and is covered by containers, cranes, and container trucks, as shown in Figure 7-5. Figure 7-5 Port area map A gantry crane in the port is responsible for hoisting containers to a specified position in the yard. A traditional gantry crane uses optical fibers or Wireless Fidelity (Wi-Fi) to connect to the central control room for remote control. Before the remote control reconstruction, the existing network of the port faces 57 Successful Applications of Network Slicing challenges such as high optical fiber costs, limited Wi-Fi coverage, difficult production network isolation, and high labor costs. With the advent of the 5G era, the port urgently needs to be upgraded to a smart one. This requires gantry cranes to be reconstructed. If HD cameras can be deployed on each gantry crane to upload HD videos to the control room, workers can view the site through HD videos and remotely control the gantry crane through low-latency channels. This resolves the above challenges while meeting the requirements of the gantry crane. To implement remote control and monitoring, the following capabilities are required: High bandwidth: Each gantry crane has 18 1080p cameras, and the required uplink bandwidth is 30 Mbit/s. Low latency: The required E2E latency is 18 ms, where the latency required for the IP network is 3 ms. High availability: 99.999% availability, less than one suspension per month. Smart Port Based on Network Slicing Remote control and remote monitoring have different network requirements. The slice management system is used to deploy low-latency and high-bandwidth slice-based private networks for remote control and remote monitoring. This provides E2E SLA assurance at different service levels, effectively meeting the remote operation requirements of gantry cranes. 58 Successful Applications of Network Slicing Figure 7-6 Smart port based on network slicing The network slicing-based smart port solution has the following characteristics: Low latency: A low-latency FlexE slice is deployed to carry control services. This implements hard pipe isolation and meets ultra-low latency requirements. High bandwidth: A high-bandwidth slice is deployed to carry video services. This meets uplink bandwidth requirements without affecting control services. High reliability: Slices support fast rerouting and service switching within 50 ms. Visualized slice SLA: IFIT is deployed to visualize the SLA of service slices, implementing fault locating within minutes. 59 Successful Applications of Network Slicing Compared with traditional ports, smart ports that adopt remote control and monitoring can cut labor costs by 75% and enable gantry cranes to move flexibly and work unattended, reducing security risks. 7.4 Smart Grid — Slice-based Power Grid Private Network Requirement Introduction China Southern Power Grid serves an area of over 1 million square kilometers, covering more than 250 million people. China Southern Power Grid built a smart grid on a communication network, leveraging sensing and measurement technologies, device technologies, control methods, and decision-making support systems to achieve reliability, security, cost-effectiveness, efficiency, and environment-friendliness targets. A smart grid involves two types of communication application scenarios: control and collection. Control services include intelligent distributed power distribution automation, demand response, and distributed energy control. Collection services include advanced metering and big video applications. Table 7-1 Electric power service scenarios Service Typical Type Scenario Scenario Description Control Intelligent Implements protection and control of the power distribution distributed network. By using automatic relay protection devices to power monitor the status of lines or devices on the power distribution distribution network, the system rapidly determines and accurately locates automation line segment or device faults, isolates the faulty segments or devices, and then restores power supply in normal areas. This scenario requires ultra-low latency and high reliability. 60 Successful Applications of Network Slicing Service Type Typical Scenario Scenario Description Demand Reduces or shifts the power load in a certain period when the response reliability of the power system is threatened, thereby ensuring the stability of the power grid. This scenario requires ultra-low latency and high reliability. Distributed Controls distributed energy forms, including solar energy energy utilization, wind energy utilization, fuel cell, and combined control cooling heating and power. Distributed energy resources are distributed on the user/load site and adjacent sites. The locations are flexible and scattered, and the number of distributed energy resources is large. This scenario features massive connections and real-time statistics collection. Collection Advanced Performs metering information in-depth of collection smart meters for to power meet consumption smart power consumption and personalized customer service requirements. In this scenario, a large amount of data is frequently collected. It features massive connections and real-time statistics collection. Big video applications Includes substation inspection robots, drone patrol for power transmission lines, comprehensive video surveillance for power distribution rooms, and mobile management and control for onsite construction activities, which require real-time video and image transmission. As such, high communication bandwidth is required. A smart grid has the following key requirements on a network: Ultra-low latency: Intelligent distributed power distribution automation and demand response services require precise control at the millisecond level. Massive connections: The power distribution network has a wide coverage, and a large number of smart meters and distributed energy resources are deployed. This requires massive connections. High bandwidth: Monitoring and inspection services require high bandwidth. 61 Successful Applications of Network Slicing High reliability: Video services require 99.9% availability, control services require 99.999% availability, and production and management services need to be isolated. Smart Grid Based on Network Slicing Smart grid services have a wide range of requirements. Slice services of different networks can specifically meet the communication and transmission requirements of a smart grid. As shown in Figure 7-7, different service scenarios correspond to different network slice types. The URLLC slice mainly includes intelligent distributed power distribution automation and demand response services. The mMTC slice mainly includes distributed energy control and advanced metering services. The eMBB slice mainly includes big video applications, such as substation inspection robots and drone patrol for power transmission lines. Multiple network slice instances can be created for each type of slice as required. Power grid enterprises can provide differentiated electric power network slice services based on the slice running status and service requirements. 62 Successful Applications of Network Slicing Figure 7-7 Smart grid based on network slicing The network slicing-based smart grid solution has the following characteristics: Intelligent slicing and guaranteed latency: MEC is deployed closer to users on demand. FlexE technology is introduced on the IP network, ensuring that its latency is less than 2 ms. Security isolation: The IP network uses FlexE to reserve resources such as physical and virtual logical resources for services in different zones of the power grid, meeting different security isolation requirements of production and management power services. High reliability of slice services: Comprehensive protection mechanisms, such as FRR, ensure high-quality bearing of power grid services. 63 Successful Applications of Network Slicing Chapter 8 Technical Prospects of Network Slicing In contrast to individual consumer services oriented to end users, industry and enterprise users require a network that can provide deterministic service assurance in terms of key indicators such as latency, jitter, and packet loss rate. This means that network capabilities need to evolve from "one pipe, best effort, and applications adapt to networks" to "differentiated connections, deterministic assurance, and network matches applications." Network slicing is the most important technical foundation for implementing the preceding capabilities. Network slicing is a landmark technology in the 5G era. It opens the door to a wide range of new industry applications and brings secure, reliable, manageable, and controllable differentiated services with committed QoS to industry markets. For carriers, network slicing helps them offer their network infrastructures to various industries in a more flexible, efficient, and open manner. In addition, slice isolation of different dimensions and degrees ensures high QoS, reliability, and security of various services. Currently, carriers are actively cooperating with industry partners, equipment vendors, and integrators to fully verify and gradually put network slicing solutions and key technologies into commercial use in multiple fields such as electric power, healthcare, industrial manufacturing, transportation, video live broadcast, and cloud gaming. 64 Technical Prospects of Network Slicing For tenants in vertical fields and third-party Over The Tops (OTTs, providing various application services for users through the Internet), network slicing allows network resources and capabilities to be used more conveniently and quickly, obtaining on-demand deterministic service assurance. Network slicing will promote in-depth cooperation between tenants in vertical fields, third-party OTTs, and carriers, and promote the emergence and development of new business models and ecological environments. Network slicing's requirements and related technologies start in the 5G era, but its applications will not be limited to 5G. The concept, architecture, and technical solutions of network slicing will be continuously verified and improved in more extensive service scenarios. In addition, the continuous deployment and application of network slicing will bring greater value to carriers, industry users, and enterprise users. 65 Technical Prospects of Network Slicing Contact Us networkinfo@huawei.com More IP Network eBooks https://e.huawei.com/en/solutions/enterprise-networks/ip-ebook 66 Technical Prospects of Network Slicing