Network slicing is one of those things 5G industry experts often cite when they’re describing the benefits of a 5G world.
At its most basic, network slicing provides for a more efficient use of the network. Instead of allocating on a serialized basis, operators can do things in parallel by allocating network resources and applying certain quality of service (QoS) elements to them. Services could be priced based on the quality or performance that a business needs; even a traditional MVNO could be structured based on slice agreements.
With network slicing and 5G, the idea is for operators’ business customers to get connectivity and data processing tailored to their specific requirements—latency and/or speed, for example. The GSMA estimates (PDF) that in combination with other enablers and capabilities, network slicing will permit operators to address a revenue opportunity worth $300 billion by 2025.
In the works for a while
Although often mentioned in the same realm as 5G, network slicing isn’t entirely unheard of in the LTE environment. The original target for network slicing, which came at the beginning of the 5G Next Generation Mobile Networks (NGMN) specification, highlighted slicing as an inherent part of 5G, but it's actually been around for years and came out of the original VPN concept, according to Sue Rudd, director of networks and service platforms at Strategy Analytics, who recently participated in a FierceMarkets webinar on the topic.
The goal is to create multiple logical networks or “slices” with specified throughput, latency and other performance parameters, often referred to as service level agreements. Each slice can support multiple sessions or flows, and slices are monitored and managed to guarantee QoS for diverse services. Ideally, all slices would share a common virtualized physical network infrastructure, but the industry isn’t there yet; today, there tend to be dedicated solutions to guarantee the QoS, she said.
As for which U.S. operator ultimately gets to 5G network slicing first, analysts give AT&T good odds because of its emphasis on network virtualization.
A key certainly is adherence to the 3GPP standards, “but it should be triggered on how well does the operator have their infrastructure technologies (SDN/NFV, etc.) in place to support all this,” said 556 Ventures principal analyst Bill Ho. “If one uses that as a marker, I’d have to say in the U.S., it’s AT&T since they’re very public and aggressive with transforming their core and adopting SDN/NFV, perhaps followed by Verizon.”
When queried by FierceWirelessTech, AT&T's vice president for mobile core and network services, Paul Greendyk, provided this statement: “Network slicing is an intriguing future tool in the toolbox for network operators. It will improve the operators’ ability to manage their network and craft solutions for customers. Many of the basic tenets of network slicing are already present in the AT&T network today, such as APN use and management.”
Based on the expected timeline for 3GPP standardization of the network slicing concepts in 5G, “network slicing will likely not be implemented widely until 2021. AT&T’s leadership in virtualizing its network is a clear enabler for network slicing, as the ability to build a software enabled ‘slice’ in the core is far easier and faster to deploy than a hardware based network,” Greendyk said.
That jibes with what Rudd and other analysts say: Virtualizing the network makes the economics far better for network slicing than a network that’s not virtualized.
Although less public about its virtualization strategy, Verizon also is seen as an early mover in the slicing space. In fact, a company spokesperson said it’s using the same core as 4G to start, and it already put a network slice in for Visible, the $40/month prepaid service that offers unlimited data at speeds up to 5 Mbps. “The standalone core will come in the 2020 time frame which will further our ability to slice the network,” the spokesperson added.
Of course, some standards related to network slicing are still underway in 3GPP, which has a lot of working groups devoted to network slicing, and Sprint is part of that process along with everyone else.
In the meantime, “we’re trying to make sure that we’re ready in every way that we can be,” said Heather Campbell, Sprint’s network director for the southern region, who has been fully immersed in network upgrades in the run-up to 5G as part of preparations for the Super Bowl in Atlanta. “Massive MIMO is our stepping stone into that,” she said, noting that Sprint recently announced a project with the city of Peachtree Corners, Georgia, which is building an intelligent vehicle test track featuring Sprint’s 5G technology combined with internet of things and micropositioning.
T-Mobile did not immediately respond to a query about its network slicing plans.
A couple years ago, network slicing was envisioned for things like video, remote surgery or cars, and at that time, Ericsson executive Ulf Ewaldsson described network slicing as running on an application-driven approach. Ewaldsson has since been hired at T-Mobile, where he’s likely going to be thinking about network slicing in the overall 5G evolution strategy.
Vendors getting ready
At Mobile World Congress 2019 in late February, it’s quite possible the industry will see some demos of the new 5G core with early pre-launch dedicated versions of 5G 3GPP network slicing, Rudd said. 5G 3GPP capabilities for a network require at least several of the new 5G core functions to be present to create even a minimal connectivity service, and most infrastructure vendors still have a massive amount of new software to develop for the new 5G core, Rudd said.
She expects, however, the big infrastructure vendors will probably be showing mostly “5G Ready” network slicing demos or LTE/4G slicing solutions. “The problem is that without the virtualization of 5G, the 4G slices are costly for operators to provide—at least for premium low latency services,” she said.
Orange, Vodafone and Verizon are all anxious to offer simple network slicing to support their SD-WAN enterprise offerings, as it already has a good business case for enterprise VPNs, especially if they can do seamless fixed and mobile access—something enterprises really want, Rudd said. Deutsche Telekom was very active in slicing at MWC 2017, but it has gone relatively quiet on slicing since then, she said. One star at MWC 2019 may be SK Telecom, which pioneered “RAN slicing” and claimed commercial network slicing in 2018. SKT also may well be one of the first to demonstrate at least a 5G 3GPP slice across the RAN, she added.
3GPP has many groups working on network slicing; the organization pushed back the full 3GPP Release 15 standalone 5G specs by three months, which means Release 16 is delayed by three months as well—now due March 2020—and that will address additional work on network slicing as well. Once the standard is finished, vendors can build their products accordingly, although the way things have been going, vendors are working on products well before the standards are solidified in part because they’re all involved in the standards process and are therefore familiar with where the specs are headed.
“From a practical point of view, I think it’s going to take some time for it to pan out,” said Chetan Sharma, founder and CEO of Chetan Sharma Consulting, of network slicing. Even if operators started implementing it, the business model still needs to get worked out, as well as use cases. Sharma said he doesn’t see any major availability of it this year, but once there’s broader 5G coverage and enterprises become aware of its advantages, it will start to hit the mainstream.
Improvements to managing congestion have been needed for some time, and Sharma noted the notion of QoS was intended as part of the LTE spec but never got used because it got ensnared in the net neutrality debate.
Some early implementations of 5G-based network slicing might consist of operators experimenting with how they manage traffic during busy times; when enterprises realize the power of slicing, it will start to take off. Then, “I could see them asking for different pricing or different models where they could have a slice … right now it’s best effort,” but it’s possible there could be a slice with 99.99% reliability and another slice that doesn’t need that requirement.
“I think the potential and the capability are there,” but operators still need to put together a business model around it, he said. In sum, some people are getting really excited and others aren’t thinking about it much. “It’s going to take some time.”