Why do we need network slicing in 5G?

Contributed by Dayan Ng, 5G Chief Architect, APAC, Juniper Networks.

Network slicing is a fundamentally new way of doing networking. Network slicing enables multiple independent end-to-end logical networks to be created on-demand and run on a shared physical infrastructure. Here end-to-end means the RAN and other access networks, the Transport and Core Networks, and the Telco Cloud, even including public cloud and enterprise on-prem data centers.

The legacy monolithic networks are best-effort where all devices, use cases and customers are treated the same. 3GPP has defined five standardized 5G network slice types in TS 23.501, namely: eMBB, URLLC, MIoT, V2X and HMTC. Each slice type has very different SLA requirements. Instead of the costly endeavour of building separate, dedicated networks for each slice, network slicing can enable and assure dedicated SLAs on the same physical network for each device, customer and use case.

Network slicing is the biggest monetization opportunity for 5G. According to global tech market advisory firm ABI Research, 5G network slicing will generate revenue in excess of US$20 billion by 2026. Network slicing enables new business models and use cases for verticals and opens new revenue opportunities for service providers.

What are the challenges of implementing Network Slicing? How much complexity does Open RAN add to network slicing?

There are multiple challenges in implementing network slicing in 5G networks.

  • Scale: With virtualization, network slicing requires orchestration across 100s to 1000s of access, edge, regional and central clouds.
  • Speed: In today’s on-demand economy, taking days or weeks to create or delete a slice is a non-starter. Slice life-cycle management requires software-driven automation to reduce time-to-market and minimize potential manual errors.
  • Service assurance: Slices are defined by SLAs. For example, extreme reliability (>99.999%) and ultra-low latency (1-10ms) are required for critical use cases. These stringent SLAs need to be delivered, measured and assured with the required QoS.
  • Security: End-to-end segmentation and security of each slice is fundamental in multi-tenancy scenarios.
  • Personalization: The network should be capable of supporting newer business models such as a network slice marketplace, where consumers and businesses can “join” a slice, “create” a slice or partially augment a network slice with their own applications. This puts additional demands on how network slice resource management is done in real-time.

Open RAN eases the complexity of implementing network slicing. O-RAN standardizes the disaggregated RAN architecture with a set of open, well-defined interfaces. This simplifies network slicing orchestration, interoperability, visibility, automation, and dynamic control, providing a clear path towards fully programmable, intelligent, and multi-vendor RAN.

What is network slicing’s role in Private 5G networks?

Private 5G networks can be deployed through different ways. You can build a standalone 5G network with its own radio and core resources dedicated for a single enterprise. You can also deliver private networks through neutral host based deployments, which provide network infrastructure as well as logical networks to private networks or verticals. Service providers can also deliver network slicing as a service to private enterprises.

It will be relatively easier to deploy private 5G networks with dedicated network resources. But the real game changer for service providers as well as industry verticals will be to implement network slicing as a service. This implementation will be the most complex but also the most lucrative for service providers and enterprises.

When can we expect to see network slicing at scale?

Network slicing is still in its early days. Service providers first need to move to 5G Standalone (SA) networks, which enable support for native network slicing.

According to a recent GSA report, 166 operators have launched 3GPP-compliant 5G mobile services. However, only 13 operators (or 8% of commercial 5G operators) have launched commercial public 5G SA networks. This means that majority of the commercial 5G networks today are non-standalone (NSA) deployments, focusing primarily on the eMBB (enhanced Mobile Broadband) use case.

Still, 75% of the respondents from the Heavy Reading 2021 5G Network & Service Strategies Operator Survey expect their company to have deployed 5G core by the end of 2022. After 5G SA is deployed at scale is when we will see service providers start to deploy network slicing. Hence, we can expect network slicing deployments by service providers with full orchestration, visibility and automation to be around 18 to 24 months away.

What is Juniper doing to deliver E2E slicing and orchestration across the RAN, Transport and Core network?

 Juniper has developed an innovative RIC platform leveraging key technology assets from the recently announced exclusive IP licensing deal with Netsia. The RIC platform extends Juniper’s vision for Network Automation, Control, Intelligence and Assurance to the RAN. As a RAN neutral vendor, Juniper’s approach is to create a RIC platform that enables powerful rApps/xApps (Juniper and 3rd party) on an O-RAN Appstore to enable new business models, personalized service experience, as well as optimize CAPEX and OPEX. Our solutions are architected as open platforms supporting open interfaces on both the north and south bound sides. This will enable easier integration with O-RAN partners in the ecosystem.

The RIC platform is also central to Juniper’s vision of building a multi-cloud, multi-domain and multi-tenant Service Management and Orchestration platform (SMO).

The Juniper SMO is built to deliver end-to-end Network Slicing with support for prescribed SLAs across the RAN, Transport and Core Networks. RIC as the RAN controller (and slice manager) complements our telco cloud and transport domain controllers (Contrail and Paragon) to provide E2E network slicing and orchestration capabilities.