Network Slicing

What Is Network Slicing?

Network slicing is an emerging network architecture that enables the creation of multiple virtual networks within a shared network infrastructure. Each virtual network is dedicated to specific service types or industry users, allowing for tailored configurations such as logical topology, SLA requirements, reliability, and security levels to meet diverse service needs. This technology empowers carriers to streamline the construction of multiple private networks, offering highly flexible network services that can be dynamically allocated and managed according to service demands. This enhances the value and monetization capabilities of carriers while facilitating the digital transformation of various industries.

How Does Network Slicing Work?

The Architecture of Network Slicing

As depicted in the diagram below, the IP transport network slicing architecture comprises three layers: the network slice forwarding layer, the network slice control layer, and the network slice management layer.

How Does Network Slicing Work?

The Architecture of Network Slicing

As depicted in the diagram below, the IP transport network slicing architecture comprises three layers: the network slice forwarding layer, the network slice control layer, and the network slice management layer.

IP Transport Network Slicing Architecture

Network Slice Forwarding Layer

The network slice forwarding layer requires flexible and precise resource reservation capabilities. It should be able to partition the physical network's forwarding resources into multiple isolated sets allocated to different network slices. This isolation can be achieved using methods like FlexE sub-interfaces, channelized sub-interfaces, and HQoS.

Network Slice Control Layer

The network slice control layer establishes distinct logical network slice instances on a physical network, configuring customized logical topology connections. It links the logical topologies of slices with the allocated network resources, forming network slices tailored to specific service requirements. This layer consists of both control and data planes. The control plane manages the distribution, collection, and computation of network slice information, while the data plane identifies and forwards network slice resources. Common technologies employed at the control layer include SRv6 and Flex-Algo.

Network Slice Management Layer

The network slice management layer oversees network slice lifecycle management tasks, encompassing planning, deployment, maintenance, and optimization of network slices.

Network Slice Address Identification

The transition to network slicing introduces a significant shift from a traditional two-dimensional network to a three-dimensional structure comprising numerous logical networks. In a conventional setup, each network node is assigned a unique IP address for identification during packet forwarding. However, this approach poses challenges in a three-dimensional network where different slices may have distinct forwarding paths and resources, necessitating the allocation of IP addresses to each node in every slice.

To address this challenge, two-dimensional address identifiers are introduced. These identifiers combine the IP address of a physical network node with a slice ID to uniquely identify a logical node within the network slice. This approach simplifies network deployment by requiring only one set of address identifiers, regardless of the number of network slices. Moreover, it eliminates the need for additional address planning or configuration during slice deployment. Additionally, the use of two-dimensional address identifiers reduces the number of routes in a slice network, enhancing support for a large number of network slices.

Two-Dimensional Address Identifier Forwarding Process of a Network Slice

In a slice ID-based network slice, a device must generate two forwarding tables. The first is a routing table, which determines the outbound interface based on the packet's destination address. The second is a slice interface's slice ID mapping table, which identifies a slice's reserved resources (such as sub-interfaces or channels) on the interface using the slice ID in the packet. When a service packet arrives at a device, the device first consults the routing table to find the outbound interface based on the destination address.

Subsequently, it searches the slice interface's slice ID mapping table using the slice ID to ascertain the reserved resources (sub-interfaces or channels) on the outbound interface. Finally, the device forwards the service packet using the appropriate sub-interface or channel.

How Are Network Slices Managed?

The network slices undergo management across four stages: planning, deployment, maintenance, and optimization.

Slice planning: In the slice planning phase, the focus is on defining the scope, bandwidth, and latency requirements based on service assurance criteria.

• Scope planning: This involves defining the extent of slicing, which can be flexible, covering the entire network, or specific areas. It addresses issues related to slice interface bandwidth utilization and network load balancing.

• Bandwidth planning: Rules are established for bandwidth allocation within slices. For shared slices, such as those in industries, the ratio of slice bandwidth to overall network bandwidth must be defined. Dedicated slices require specific bandwidth allocation.

• Latency planning: Network latency parameters are specified to ensure the defined range meets service requirements.

Slice deployment: A controller is utilized to deploy slice instances, which includes creating slice interfaces and configuring parameters such as bandwidth, VPNs, and tunnels.

• Creation of a network slice on the controller: Slice interface types can vary from physical interfaces to FlexE interfaces or channelized sub-interfaces.

• Activation of the network slice to generate basic configurations, including setting up the slice interface's IP address and enabling Interior Gateway Protocol (IGP). The device reports the slice's Layer 3 topology to the controller using Border Gateway Protocol-Link State (BGP-LS).

• Deployment of an SRv6 path within the slice.

• Implementation of VPNs (e.g., L3VPN and EVPN L2VPN) within the slice.

Slice maintenance: The controller utilizes technologies like iFIT for monitoring service latency and packet loss. Telemetry technology is employed to report network slice traffic volume, link status, and service quality information, providing real-time visualization of network slice status.

• Slice visualization: Monitors network slice traffic volume, link status, and service quality, offering comprehensive network status display and network slice profiling.

• Fault diagnosis and prediction: Real-time monitoring of network slice status enables proactive analysis of root causes of network faults and prediction of potential network issues.

• Fault rectification: Automatically takes corrective measures, such as path adjustments and optimizations, to address traffic congestion and fault scenarios.

Slice optimization: Seeks the optimal balance between slice network performance and network costs based on SLA requirements. This optimization involves bandwidth optimization and slice capacity expansion.

• Bandwidth optimization: Utilized when there are sufficient resources within a slice but inadequate partial bandwidth resources, ensuring bandwidth requirements are met.

• Slice capacity expansion: Implemented when a slice is overloaded and service bandwidth cannot be assured through optimization, requiring expansion of slice capacity.

Why Network Slicing Matters？

In the realm of cellular communications, network slicing empowers businesses to finely control traffic resources. Each traffic slice can have its own distinct resource requirements, Quality of Service (QoS) parameters, security configurations, and latency specifications. For instance, a network slice supporting high-definition streaming video would differ significantly from one used to monitor an Internet of Things (IoT) lighting system.

In conventional, non-sliced networks, devices often access more resources than necessary. For example, an employee's cell phone may not require 100 Mbps to send a simple message via an app. Network slicing conserves resources by considering the context and specific use case of each application, then allocating the appropriate amount of resources accordingly.

With the advent of new core network technologies such as NFV, implementing network slicing has become more feasible, particularly over 5G networks. Enterprises, mobile network operators, and managed service providers all stand to gain substantial benefits from the adoption of network slicing.

The Benefits of Network Slicing

The advantages of IP transport network slices are primarily demonstrated in four aspects: resource and security isolation, deterministic latency, customizable flexible topology connections, and automated slice management.

Resource and Security Isolation

Network slice isolation serves the purpose of maintaining service quality by preventing any service disruptions or abnormal traffic within a slice from impacting other slices on the same network. This ensures that services within different network slices remain unaffected by each other. Particularly critical for services with stringent latency and jitter requirements, such as smart grid, smart healthcare, and smart port applications, maintaining isolation is essential to prevent performance degradation due to interference from other services. From a security standpoint, it's imperative to ensure that information pertaining to services (e.g., finance and government services) or users within a network slice remains inaccessible to users in other slices. Effective security measures are thus necessary to enforce isolation between different slices.

IP transport network slices offer three levels of isolation based on the extent of separation: service, resource, and O&M isolation.

Service isolation ensures that service packets within one network slice are not routed to service nodes within another slice on the same network. Essentially, service connections are segregated between different network slices, rendering services from different tenants invisible to each other within the same network.

Resource isolation entails segregating the network resources utilized by one network slice from those of another. This can be achieved through either hard isolation or soft isolation, each differing in the degree of separation. Hard isolation guarantees exclusive network resources for each slice, effectively preventing any interference between services. On the other hand, soft isolation allows slices to utilize both dedicated resources and shared resources with other network slices. This hybrid approach enables services to be somewhat isolated while retaining certain statistical multiplexing capabilities. By employing a combination of hard and soft isolation, carriers can tailor network slices to meet specific resource requirements, allowing a single physical network to accommodate diverse service-level agreements (SLAs).

O&M isolation addresses the need for independent management and maintenance of network slices allocated by carriers, akin to using private networks. Network slicing achieves O&M isolation through the openness of interfaces on the management plane.

Deterministic Latency

Various services exhibit diverse requirements concerning bandwidth and latency. Traditional services typically tolerate end-to-end network latency exceeding 100 ms. Conversely, real-time interactive and industrial control services, such as power grid differential protection, demand ultra-low latency of 2 ms in an IP transport network. Furthermore, these services necessitate deterministic and committed latency assurances. Network slicing facilitates the deployment of different services across distinct slices, thereby ensuring deterministic latency for interactive and control services.

Flexible Topology Connection Customization

As services and traffic patterns evolve to encompass multiple directions, network connections become more flexible, intricate, and dynamic.

Network slicing offers the capability to customize logical network topologies and connections tailored to diverse industries, services, or users, addressing varied network connection requirements. Users within a network slice only need to be familiar with the slice's logical topology and connections, rather than the entire topology of the underlying network. Moreover, services within a network slice are confined to deployment within the corresponding slice's topology. This simplifies the network information required for users of network slices. For carriers, this prevents excessive exposure of internal information from the underlying networks to network slice users, thereby enhancing network security.

Automated Slice Management

With the continuous expansion of service types and scales, network management complexity escalates rapidly. Manual network management is no longer a viable option. Instead, dynamic and efficient network management necessitates the adoption of automated technologies.

The network slice manager offers comprehensive lifecycle management for network slices. It streamlines the entire process from user intent to service provisioning, encompassing network slice planning and deployment, flexible mapping of services to slices, real-time visualization of slice services, and dynamic adjustment and optimization of slices. Additionally, it provides tenant-level refined service management.

As network management automation advances, intelligent technologies may become extensively utilized in each phase of network slice management to enable intelligent network management.