Network Slicing for AN Resource Sharing and Rapid Service Development

Release Date:2019-09-16 By Yu Pingzhi Click :

 

After a decade of rapid development, the scope of optical broadband services has gradually expanded from traditional home broadband access to multi-ISP/ICP access, wholesale or multi-tenant access, enterprise private line access, and mobile base station transport. Traditionally, access networks (ANs) are separately built and independently operated by different operators or service departments, which greatly increases the construction period and cost. A paradigm shift is needed to allow AN infrastructure to be shared to boost utilization and reduce cost. Against this background, network slicing is born.

In a telecom network, network slicing is an innovative technology used to logically divide a physical network and virtualize multiple logical devices on one physical device. The purpose is to cut overlapping investment while achieving service convergence, security isolation and differentiated services. In an optical access network, the slicing technology is mainly applied to optical line terminals (OLTs). By splitting a physical PON OLT into multiple logical PON OLTs, key network resources including access offices (AOs), OLTs and optical distribution networks (ODNs) can be shared so that services of different types and for different customers are separately operated and maintained to maximize investment value.

Application Scenarios  

The main application scenarios of network slicing are multi-service shared access and multi-operator shared access.

Multi-Service Shared Access

An operator can use one OLT in an area to connect multiple service types, like fixed home broadband, enterprise private line and 4G/5G base station transport.

The operator can select a proper slice granularity according to the service operation mode, such as by upstream port, PON card, PON port, and ONU or in a mixed mode. It can also opt for slicing by upstream port, PON card, or PON port to meet the operation and management (O&M) requirements of different services. In this way, the operation support systems (OSSs) of different business departments can be independently managed to simplify the OSS development and interconnection processes including automatic service provisioning and O&M.

Multi-Operator Shared Access  

One OLT in an area can provide multiple operators with access services (Fig. 1). The operator owning the OLT is called the infrastructure provider (InP), and the operators sharing the OLT for service development are dubbed virtual network operators (VNOs). Based on its business model with the VNOs, the InP chooses a slice granularity, logically splits the OLT, and delivers the specified resources to the VNOs for independent operation.


To simplify O&M, the InP can use the slice granularity of PON card or PON port to interconnect the OSSs of VNOs for automatic service provisioning and O&M as well as sharing OLT, AO or some pipe resources.

To maximally share AN infrastructure resources, the InP can employ the slice granularity of ONU level to interconnect the OSSs of VNOs for automatic service provisioning and O&M as well as sharing more key resources such as optical fiber pipes and ODNs in addition to OLT and AO. This operation mode allows the InP to maximize its investment value while letting users choose broadband access services of a preferred operator.

Implementation Scheme

Network slicing creates multiple mutually isolated slices on one physical OLT. With the technology, resources can be allocated, reused and recycled on demand. Services can be independently deployed, operated and maintained on the slices. As a result, AN resources can be shared.

On-Demand Resource Allocation and Reuse   

An OLT can be divided into physical and logical resources. The physical resource includes upstream ports, PON cards, PON ports, P2P ports, and ONUs, while the logical resource contains hardware forwarding entries such as VLAN, multicast, MAC and ACL. These OLT resources can be allocated on demand and then exclusively used by the intended slices to prevent preemption. In the early stage of service rollout, OLT owners can allocate resources according to their service plan. In the later stage, they may expand or release the resources, depending on how well the services are developed.

A network identifier, such as a VLAN ID or an IP address, can be reused by different slices. This enables business departments to plan independently without causing resource conflicts. Different slices can also use the same IP address to manage different devices.

Independent and Isolated Management, Control and Forwarding Planes

After a physical OLT is sliced, each slice has its own upstream port and user port resources. To fully isolate the services of the slices, the management, control and forwarding planes of each slice are independent of each other. This is actually an instantiation of functional entities related to the physical OLT, with different instances horizontally isolated from each other. Each slice also has its exclusive software resources that are abstracted by the operating system, including CPU, process, database and file resources.
—Management plane: Each slice has an independent in-band management channel, management IP address, system MAC address, and configuration files, as well as its own SNMP, CLI, FTP, NTP, and NETCONF protocol entities.
—Control plane: Each slice has independent IGMP, DHCP, and dynamic routing protocols.
—Forwarding plane: Each slice has independent L2 and L3 forwarding engines and is configured with independent VLAN and MAC tables.

As a multi-instantiation of the management, control and forwarding planes of OLT significantly increases computing and storage overheads, OLT not only needs to support multi-instance slicing on the platform architecture, but also needs large-capacity memory and high CPU performance. Therefore, new-generation OLTs in the industry can basically offer slicing functionality.

Traditional service isolation technologies like VLAN and VPN just create logical channels in a physical OLT. By contrast, the slicing technology completely isolates the forwarding and control channels of different slice domains. Network attacks and device faults, such as MAC flooding, in one slice domain do not affect the other slice domains. Isolation between the management, control and forwarding planes of different slice domains is quasi-physical and therefore affords higher security. Within a slice domain, traditional VLAN and VPN can still be used to implement link-level security isolation.

Deployment Suggestions

Operators need to deploy, operate and maintain network slices in the same way as they treat physical OLTs. They need to interconnect the slices with OSS to carry out service provisioning, alarm processing and trouble ticket dispatching. These requirements are the key issues and major difficulties that must be addressed when applying the slicing technology to the existing network. What OSS functions should be developed depends on the management needs of the network. The original processes of the OLT should be minimally impacted by the development.

Selecting Appropriate IP Mode to Manage Network Slices

Most network slices can be managed through multiple or one IP address. In the multi-IP mode, the element management system (EMS) uses different IP addresses to manage different slices. It is recommended to use the multi-IP mode to add and maintain IP addresses of the slices in the resource management system, which can minimize the impact on the OSS process.

Planning and Deploying Resources in Advance

According to the business development mode, the operator should select an appropriate slice granularity, plan the physical and logical resources of the slices, and synchronize the physical resources to the OSS resource management system. If the existing network needs end-to-end slice management, the operator should develop slice management specifications for northbound instructions.

Choosing Appropriate Alarm Driver and Performance Collection Mode

In the physical OLT, alarms of all the slices can be viewed and operated, while in a slice, only alarms generated under the specific slice can be viewed and operated. In actual applications, depending on the management requirements, alarms of the physical OLT, network slice, or both can be used to dispatch trouble tickets.

ZTE's flagship next-generation optical access platform, TITAN, has diverse network slicing capabilities. With the slice granularity by upstream port, PON card, PON port and ONU or in a mixed mode, TITAN allows for access to multiple services like home broadband, multi-ISP/ICP, wholesale or multi-tenant, enterprise broadband, and mobile transport.