Alla Goldner: Director, Technology, Strategy & Standardisation at Amdocs
1. Network Functions Placement is essential to running virtual and cloud-based networks
As operators deploy virtual and cloud native functions, there is a need to manage and optimise those functions to avoid resource duplication and inefficiencies. Network Functions Placement (NF Placement) is a set of smart algorithms which deliver an input into the network management function so that it picks up the right network functions according to the specific requirements of a service. This includes mapping the requirements to the availability of the network, invoking the right existing physical functions, as well as new virtual and cloud-native functions depending on the required parameters for latency, bandwidth and so on. This whole process is called NF Placement.
A potential use case that can benefit from NF Placement is providing coverage and capacity for enhanced real-time video services during a stadium event. During the event the service provider can allocate network functions, radio resources, mobility management, media processing and so on, closer to the stadium in order to meet a spike in demand during the event. But when the event is over and demand subsides, the NF Placement orchestrator moves those resources back to a more centralised function. That creates a very dynamic and cost-efficient allocation of resources.
Other uses might include meeting on-demand requests for instant high bandwidth from smart CCTV systems, or for ultra-low latency from real-time applications such as AR video conferencing.
2. NF Placement will be required to operate 5G networks at scale
There are potential interim workarounds to NF Placement – but they won’t work at scale. One alternative is to try for a more static workaround, where resources are replicated across the network to meet potential demand and peak service level requirements. However, once 5G networks are deployed at scale, their size and complexity will mean that network function duplication will simply no longer provide a valid workaround. This is clearly going to be inefficient and will be almost impossible to structure for services that are very dynamic in terms of their timing or mobility.
3. NF Placement is about more than geography
“Placement” sounds like we are dealing specifically with “where” to invoke functions, and of course geography is an important input parameter, especially for latency, but it is not the only consideration. NF Placement is aware of network topology and locations, but it is also aware of service, slice and application requirements. There might be a preference for which cloud a function should reside on – public or private – and that will have meaning in terms of the Service Level Agreement (SLA) requirement. There might be specific values for resiliency, or security and scalability, which those network functions need to satisfy for the service. So all of these requirements are selected together and then combined in an algorithm that takes into account the business intent of the operator.
4. NF Placement requires an end-to-end view
Building this requires a multi-vendor view across RAN, transport and core that is adaptive in real-time. That in turn requires an Operations Support Systems (OSS) and Network Management System (NMS) that can map functions to requirements. The OSS & NMS know the location and types of network functions as well as their capabilities such as scaling and self-healing, so that they can map the service requirements to function capabilities in order to satisfy those requirements. Then the OSS maintains the service, getting closed-loop measurements, and is aware when the functions need to be scaled up or down, or moved from a public or a private cloud.
5. NF Placement is key to slicing, which is key to monetisation
5G provides the impetus, and in fact necessity, to build NF Placement capability in the network. By constructing a 5G network, you are building in new capabilities to provide differentiated services that can be guaranteed with SLAs and managed by a Network Slicing Function within the new 5G network.
5G won’t be properly monetised without network slicing and NF Placement is key to giving service providers the ability to manage network slices end-to-end. CSPs really need NF Placement to pick the right network functions each time a service is run and to be able to change those functions dynamically. Slicing is really where NF Placement comes into play.
To guarantee those SLA commitments that CSPs offer around network slices, there is a need for dynamic, adaptive, real-time NF placement that optimises network resources.
6. Building out NF Placement capability is a three-stage process
The way we see it is that operators will move through three stages, from the more static, pre-instantiated NF Placement of today, to something much more adaptive, and then finally to fully autonomous NF Placement.
Today’s static methodology involves planning resource usage based on peak loads, and using workarounds such as resource duplication or hardware optimisation techniques to mitigate against over-provisioning. During this phase, network functions are placed once only, and the orchestrator uses load balancers and policy rules to split the load on the network between available resources in real time.
As 5G establishes a more disaggregated architecture, formed of VNFs and CNFs that can enable dynamic network slicing, that will bring a move to the second phase, ‘Adaptive NF Placement’. In this phase AI and ML assist in identifying slice utilisation patterns and alert the NF Placement orchestrator of thresholds crossed before actual QoS degradation occurs.
The adaptive nature can also be implemented with ‘semi-automatic’ NF placement, where the NF placement orchestrator chooses from two or three pre-instantiated locations in real time, based on predefined policy rules and real-time AI/ML.
The final stage, fully autonomous NF Placement, demands that the NF placement orchestrator handles the entire lifecycle of the service or slice with its associated NF resources, using AI/ML as needed to fine-tune dynamic placement. Here, the ability to dynamically orchestrate and re-allocate Network Functions is key to further optimising and improving real-time placement over private and public clouds and even over multi-operator networks as required.
7. Service Providers can begin today to build the key capabilities to enable NF Placement
CSPs need to start thinking about how they can enable their NMS and OSS to bridge between the network layer and overlying business systems.
That requires investment in an intelligent, intent-based NF Placement solution that takes the business inputs that define a differentiated service and translates them into the underlying parameters for the orchestration system.
If service providers build this real-time, dynamic, end-to-end network management and orchestration system, they can deliver on the true promise and potential of 5G.
About the author:
Alla Goldner is Director, Technology, Strategy & Standardisation at Amdocs
