Can You Cut it?
It’s hard to find any unused superlatives to describe the leap in connectivity, capacity, reliability, latency, and the sheer diversity of services that 5G will bring. Let’s all agree on ‘monstrously better’. But there is a catch: 5G is also monstrously complex. The one area that seems to cause more puzzlement than, say, multi-edge access computing, Xhaul, or synchronization, is network slicing.
So long, ‘best-effort’
If you need a reminder of 5G’s new whizz-bang services, this article gives you a good introduction. The upshot? We can kiss goodbye to ‘best-effort’ mobile broadband, and instead, pucker up to a raft of exciting new 5G services. Lots of sectors and areas are licking their lips at the prospect – from media to education to healthcare to driverless cars to IOT to mission-critical industries to smart cities.
You can see why. Ericsson and Arthur D Little (2017) estimate the 5G-enabled industry digitalization revenues for ICT players will reach US$1.3 trillion by 2026. With, enterprise seen as the biggest opportunity ($300 billion expected revenues by 2025. Source: GSMA, 2016). Now back to network slicing…
So what is network slicing and why the fuss?
Put simply, network slicing allows you to cut or slice one physical infrastructure into lots of virtual networks, which you can customise to meet the specific needs of the applications, services and resources you want to run over it.
This side-steps the need to build different networks for each service you want to offer, which would be hugely expensive. Instead, you have one set of network resources, serving multiple slices. And you allocate a slice for each service type – so ultra-high capacity, or ultra-high reliability, or ultra-low latency, for example. And possibly, one or two sub-virtual networks to manage different types of services/customers within a slice.
What does a slice look like?
Imagine a network as a pool of assets – so PNFs (Physical Network Functions), VNFs (Virtual Network Functions), connectivity, bandwidth and compute. You bring these together in different combinations to create a slice with a specific set of operational parameters. Each slice supports a separate control plane and user plane. In addition, each slice is dynamic enough to allow you to dial up or down extra capacity, processing and VNFs as service requirements change. As you can imagine, all this requires clever orchestration of an already complex system.
So what slice-enabling technologies are there to support services with wildly different requirements on the same network? Basically, there are four key ones to consider, depending on your end-goal:
- OTN or wavelength slicing
Use this to create network slices directly across the optical transmission network. And use router bypassing to provide low latency and free up router ports for services travelling from edge to core. OTN is important because it is low cost, deterministic, and uses the optical spectrum efficiently.
- Segment Routing (SR) and enhanced VPN (eVPN)
As an enabler, segment routing adds determinism to packet networks, guaranteeing latency and reliability. Source-based routing, controlled by a centralized PCE, makes for better planning of precise, deterministic paths across the network and offers full sight of all transport resources. By combining them, you can create network slices that meet the specific service demands – such as low latency or high capacity – of the service type being transported.
- Flexible Ethernet (FlexE)
FlexE’s deterministic transport removes flow control when mapping IP packet services onto the optical layer. Acting like a clutch between IP and OTN slices, it gives you more efficient dynamic bandwidth capacity, but little or no latency associated with flow control. FlexE is the link between specified network assets for each slice you need.
You won’t need slicing for 5G-enhanced mobile broadband. But, after enhanced broadband, companies will look to make revenues from new services like autonomous vehicles and remote surgery. These will need ultra-reliable low-latency connectivity (URLLC) slices, along with segment routing. So you will need to make sure your 5G network can easily evolve to support segment routing, FlexE and OTN, for those new revenue-generating services.
At the end of the day…
Ultimately, 5G’s success will depend on its ability to support diverse services with diverse latency, capacity and availability requirements. Network slicing, working alongside other network assets, will be essential for any mobile network operator keen to cut themselves a generous helping – or slice – of what looks to be a very large pie.
To learn more about ECI’s 5G Solutions, click here.