Partner Content - Scaling up down under: Wholesale broadband at hyperscale

NBN Co is Australia’s main broadband wholesale provider, connecting homes and businesses across a continent of more than seven million square kilometres.

6 Min Read

NBN Co is Australia’s main broadband wholesale provider, connecting homes and businesses across a continent of more than seven million square kilometres. Today, our network connects 20 million people across a footprint of over 12 million premises and carries over 80% of the nation’s download traffic. But we’re not stopping there - we intend to increase the reach, capacity and capability of our network.

This is hugely important in the face of Australia’s insatiable hunger for data. In the last 12 months, we’ve seen network traffic usage records be set and exceeded four times. The average nbn user downloads around 443 Gigabytes a month, and around 10% of our active users download 1TB or more on a monthly basis.

Half of our network traffic comes from less than 20% of services, with real time entertainment being the driver of consumption. And while some usage patterns remain constant over the years, such as 9pm peak usage time, we’re also seeing new drivers like online gaming influencing our network, driving huge traffic spikes.

We recorded our largest network traffic day in December 2023 – when a Fortnite update saw network traffic rates peak at 29.85 terabits per second.

Keeping pace with this growth now and into the future requires us to scale the entire nbn network - from Fixed Access to core aggregation and transport - which helps minimise any bottlenecks from users to our retail service providers (RSPs).

As well as enabling more than five million new homes and businesses to become eligible to upgrade to Fibre to the Premises (FTTP) connections by 2025, the extension of our FTTP footprint allows us to consider the introduction of the next generation of bandwidth (XGS-PON) for greater than 1 Gbps services in the future, the deployment of Distributed Access Architecture (DAA) to scale our Hybrid Fibre Coaxial (HFC) access in coming years, and the roll out 5G mmWave technology across our dedicated fixed wireless access network to deliver faster speeds and enhanced coverage for regional Australia.

These changes are on a lot of global network operators’ “To Do” lists, however, contributing to our ambition is to achieve “wholesale at hyperscale”, and, for this, we turn our attention to the tech world – where hyperscale is foundational.

Having spent a large part of my career at both telco and tech companies, it has become apparent that while requirements at a tech and telco can be substantially different – there are best practice approaches in both industries that when adopted can enable a wholesale network to achieve hyperscale.

Let’s look at some of these best practices across three key factors: network architecture, hardware, and software.

Typical telco architecture is based on thousands of nodes in a region, built over decades of history bringing multiple technologies into play. In terms of hardware, there are millions of commodity devices across customers’ premises – which must support customer needs, product roadmaps and network features. Customers want these features within a form factor that meets the functionality and aesthetic for a modern home. Telcos need to meet these expectations, at massive scale, within commodity device economics. On the software side, due to the extensive history of the network and multiple generations of hardware, if the software is left unchecked it will grow into a jungle of monolithic and segmented proprietary systems and data.

In contrast, a tech company’s architecture will have fewer nodes, each with much higher capacity. This comes with its own constraint challenges, the prime being able to find enough power for such concentrated compute regions. Tech companies will typically have built their network more recently and thus there will be less variance in their hardware. The focus then is on remaining ruthless in removing variants as their network ages. This homogenous network has enabled tech companies to virtualise much of their core.

With this software enabled network, tech companies have moved towards a service-driven architecture - where reusable services are used to build new products or new network functions, speeding up development time and simplifying integration and operations.

Combining these best practice approaches, we know that our architecture must always support a massive amount of nodes, however, we can be stricter on the technology variants, drive towards the virtualisation of network functions and embrace the adoption of a service-driven architecture. Doing so would support our drive towards using open standards and frameworks to build new functions and product features rapidly by re-using software functional blocks and standard interfaces.

We can’t change the commodity economics of customer premises equipment: our priority is customer experience, therefore, as an industry we must keep driving innovation to build these enhanced functions and capabilities within economies of scale.

The final component to achieving hyperscale is automation. Both telco and tech companies are driving and embedding automation to deliver key business outcomes like operational and cost efficiencies, scalability, and importantly a better customer experience.

There are five key enablers to achieve successful automation:

  1. Structured, accurate data in a standard model (open and not proprietary) is vital for getting visibility of the network to view network health and identify network issues in real time.

  2. Network Abstraction separates network resources from the services it supports and enables us to grow and develop each function independently.

  3. Artificial Intelligence and Machine Learning have immediate use cases in detecting complex or silent failures, as they enable us to move towards the panacea of predictive and non-customer impacting maintenance.

  4. Open APIs drives simplicity throughout the software stack resulting in fast development of new features through the re-use of existing functions and tools and supporting multi-vendor interworking

  5. The most fundamental enabler, though, is always understanding what you are automating. We must keep simplifying processes and ensuring we have the right mix of skills within a skilled workforce.

So how are we going about things at NBN? In FTTP, we are the first in the southern hemisphere to deploy the Nokia MF14 OLT, and we’ve recently completed a successful technical field trial in our live network of 25G PON and other multi-gigabit technologies.

This demonstrates the opportunity for the NBN network to offer future multi-gigabit services, aimed at meeting the demands of Australian homes and businesses well into the future - for both retail providers and housing developers who choose NBN.

For our HFC network, we will look to industry direction, where using a Distributed Access Architecture to bring fibre deeper into the network is an efficient way to scale capacity to meet this exponential demand.

And a final crucial component is to ensure we keep ahead of the curve through partnerships between industry and academia such as our ASTRID program.

And fundamental to this evolution is the fibre itself. Our Network now extends to over 300,000km of fibre optic cable, and with the Fibre Connect Program we're growing this by an additional 70,000km across towns and communities to access full fibre, that’s deploying at a rate of 500km per week! 

This is how NBN is using global hyperscale approaches to support Australia’s exponential growth in data. With this investment, we are aiming to make more than 10 million premises gigabit-capable by 2025: that’s almost 90% of Australia’s homes and businesses. And with it comes the possibilities as endless as the Australian outback.

About the Author(s)

Partner Content by Andrew Leong

NBN Co.

Andrew joined NBN in Dec 2021 and leads the key functions that bring together the technology and architecture decisions for the future of the NBN network.

Prior to this, Andrew was based in Silicon Valley, California for 6 years, where he was Meta's Global Group Engineering manager responsible for leading the research, design and delivery of highly scalable and reliable backbone network solutions to connect Facebook's Datacenters to over 4 Billion global users.

Andrew brings extensive experience in scaling and optimising complex networks and systems through automation, webscale software solutions and deep technology expertise.

Andrew has degrees in Electrical Engineering (Hon) and Finance from the University of Melbourne, Australia and is constantly active in driving industry research and has even achieved a Guinness World Record for building the longest terrestrial optical link.

Subscribe and receive the latest news from the industry.
Join 62,000+ members. Yes it's completely free.

You May Also Like