With COVID19 having changed the way we live, work and educate, there is a growing demand for new 5G services. Enterprises are wanting to operate at greater levels of scale and automation, and with remotely working employees. Consumers are accessing a broad range of digital services from home and will want the same capabilities as they start to move outside of the home . This puts increased pressure on service providers to make 5G available across all markets and find a way to monetize these new services. Simply put, 5G’s enhanced capabilities - speed, capacity, latency, battery life, and volume of connected devices open a world of possibility for new use cases. The true potential and benefits of 5G are its ability to enable, deliver, measure and charge for new and differentiated services.
But the 5G network cannot just do this on its own. Differentiated services introduce a significant level of complexity as the network needs to allocate resources and capacity to deliver the unique QoS requirements of each service. A next generation of OSS software is needed to operate and automate the network and deliver the differentiated services. Two of these new software capabilities are network function placement and network slicing.
Network slicing allows a communication service provider (CSP) to allocate the network for different purposes – effectively enabling one network to act as many different types of networks. The introduction of network slicing makes the separation of different traffic types possible, allowing for functional differences between the different slices . Another byproduct of the creation of slices is the potential for entirely different business models between the types of slices created.
Network Function (NF) placement is perhaps one of the most critical parts of the 5G implementation process. Static placement can cause issues later on when providers try to meet service level agreements (SLAs) for different needs.
“Once upon a time, you could think of a network as a physical box, sitting in a physical location, doing a pre-determined job. As part of the network planning and design process the location and capability and capacity of the box would be decided and once deployed it would not move. As networks become software — they’re becoming virtualized, they’re becoming cloud native — they no longer have to live in a single physical box in a single physical location and they no longer have the same capacity limitations,” says Angela Logothetis, Chief Technology Officer of Amdocs Open Network. “That really gives you the ability to be able to determine what is the best place to put this function based on which function it is, what it has to do and the SLAs it needs to deliver. Network placement becomes a key capability that lets you deliver those SLAs. You might have robotics in a factory that need to work together in-real-time so that the production line keeps going. When the service provider is designing that service for the customer, they need to place the functions as close as possible to the factory to deliver the required latency.”
Another topical use case is for firefighters combatting wildfires in nations like Australia and the United States.
“There is a real benefit in being able to dynamically place resources and capacity to support first responders and emergency services. Natural disasters can damage or destroy infrastructure right at the time it is needed most and there’s a real need to be able dynamically rebuild,” says Logothetis. “And actually you can get really smart about prioritizing what you’re going to build and when, and not just be saying, ‘This location is damaged, so I’m going to dynamically build new capacity here, but also be saying the fire service is here, and now they are moving over here, so we’re going to dynamically build capacity and coverage around the fire service as they move to fight these terrible fires.’ That’s where dynamic placement really comes into play.”
Other use cases that require network function placement include:
On-demand applications that need instant high capacity, like smart city camera surveillance systems.
Real-time applications that require ultra-low latency, like a 3D video augmented reality video conference.
Mobile and nomadic users that increase the geographic load volume, like those who attend stadium events. Network functions and workloads can be placed closer to the stadium’s geographic location for the duration of the event to reduce latency.
These are some potential use cases that highlight why CSPs should think ahead when it comes to network function placement and how this stands to affect later stages of 5G implementation. Although replicating network resources at the edge of the network would solve some problems in the short term, homing and placing network functions in all the edge data centers would create redundant, inefficient and costly resources. Indeed, once 5G networks are deployed at scale, their sheer size and complexity mean that network function duplication will simply no longer provide a valid workaround.
“Network virtualization enables a service provider to be very agile and responsive in what capabilities and what capacity they place where in the network. It comes back to having the ability through software to bring up network infrastructure and then having the intelligence around that software to say where am I going to bring it up and how does it actually need to work?” says Logothetis.
The recommended approach to dynamic network function placement is in three stages, with each stage laying an important foundation for the eventual result.
In today’s homing and NF placement systems, orchestrators rely on workarounds, such as resource duplication and hardware optimization techniques, to inform NF placement. NF resource planning during this phase is based on peak-hour load and demand. Network functions are placed once only, and the orchestrator uses load balancers and basic policy rules to split the load on the network between available resources in real time. Meanwhile, service providers use their own resources at the edge data center or in some cases utilize public cloud resources to avoid making a significant upfront investment in data centers.
The second phase comes with the introduction of 5G disaggregated architecture and the dynamic nature of scalable virtual and cloud network functions (VNFs and CNFs). During this phase, network slicing acts as a critical enabler to the operation and automation of a network that offers differentiated services and charges for those services. Artificial intelligence and machine learning (AI/ML) assist in identifying slice utilization patterns and alert the NF placement orchestrator of thresholds crossed before degradation of the service provided occurs.
“Standalone 5G brings into play that new 5G core, and that’s where some of this capability like slicing comes into play. That’s all controlled out of the standalone core. Some of the many things that we are actually looking forward to being able to get out of 5G, like being able to put network function at the edge, are powered by this core,” says Logothetis.
“It’s not just about what hardware or cloud the clients have around this network function; it’s also geographically where is the best location to place it. That may be out near the customer, it may be at the edge of the network, it may be at the center of the network. These are some of the placement decisions that service providers will be able to make as they move from non-standalone network into standalone network and into edge. They are not one-off decisions because these networks are dynamic and constantly shifting. It’s actually not possible for a human to sit there and be making these decisions — you need software to do it. “
The second phase could be implemented with “semi-automatic” NF placement, where the orchestrator chooses from two or three pre-instantiated locations in real time. A predefined and monitored environment enables a higher level of placement autonomy based on a pre-measured set of network capabilities and policy rules, but it still requires operator intervention to ensure the autonomous functions work as intended. At the stage of adaptive NF placement, the orchestrator must be able to support closed-loop operations, using feedback from telemetry and KPIs to support key placement decisions made in real time. Where necessary, network functions can then be reallocated to maintain the guaranteed QoS and committed SLAs for each service or slice.
The third and final phase is fully autonomous NF placement, where the NF placement orchestrator handles the entire lifecycle of the service or slice with its associated NF resources. At this stage, artificial intelligence can be used as needed to fine-tune dynamic placement and make critical decisions about resource allocation. The ability to dynamically orchestrate and re-allocate NFs is key to further optimizing and improving real-time placement over private and public clouds and eventually over multi-operator networks, as needed.
“Slice management enables the service provider to define in a very intuitive way the rules around slicing, and for the slice manager to actually implement those on the network and then monitor it and say, ‘Ok, the way I’ve built the network for this slice, is it really delivering the quality of service that I promised for this slice?’ If it’s not, then it re-implements the slice or reallocates network capacity,” says Logothetis.
Amdocs 5G Slice Manager helps CSPs address operational challenges that naturally arise from implementing network slicing and monetizing the result. With end-to-end lifecycle automation for cross-domain, multi-vendor network slicing, it becomes much easier for CSPs to address the significant complexities that can arise .
“In terms of slice management, the first thing we see CSPs starting to do is to define the different ways they want the network to be used. They may say, ‘Here is a slice of my network that knows how to protect the battery life of a device,’ so we can think about baggage tracking or parcel tracking. We can have tiny little sensors inside the packages, and a tiny sensor means a tiny battery, so the key thing is how do we protect the battery life of that device. So they’ll put up a slice of the network that they’ll pull all self-sensors onto, in order to protect battery life. They may do another slice for latency and another slice for the security, and so they will set up these slices based on a certain characteristic of 5G.”
While CSPs may wonder where to start with placement of functions, slicing of network and monetization of differentiated services, Logothetis says the key is start somewhere.
“There is so much that is new. It’s not just the network architecture changing, but it’s edge coming into play, new services, and new experiences, and new monetization. It’s a fundamental shift in our industry. We get a lot of questions about how do CSPs go about doing it,” she says. “Our simple answer is, it’s a bit like the old joke: How do you eat an elephant? It’s one bite at a time. The key thing with all these things is to get started. Pick the right infrastructure. Actually, service providers are very good at doing that. They’re good at picking the right architecture, the right infrastructure. Pick that early use case that will not only move into slicing and placement and monetization of 5G, but will also help them build the level of innovation in their company to take these things forward. Because our software, like much of the network software, is now available in the cloud, it is actually quite quick to get up and running, and to be able to build early use cases, bring them to market, scale the cases that succeed, and fast-fail the ones that don’t.
As CSPs build 5G networks and evolve from 5G non-standalone to 5G standalone, they should at the very same time, be transforming their OSS and orchestration software to enable dynamic network function placement and network slicing. With these capabilities in place along with convergent charging SPs will be able to take advantage of 5G network capabilities