Virtualization

5G network slicing is promising – but it isn’t easy

5G promises to let communications service providers (CSPs) ‘slice’ the network to accommodate a wide range of applications with very different reliability and throughput requirements. An IoT sensor network, for example, requires relatively little bandwidth – sensors need to transmit small amounts of data sporadically but frequently – but an application like remote surgery needs huge amounts of bandwidth, extremely low latency and guarantees of availability and reliability.

To accommodate such wide-ranging services, CSPs envision turning the network into a multi-tenant, multi-operator platform, on top of which they’ll create individual network slices. These slices will be orchestrated and assured end to end through automated closed control loops using intent-based management, analytics and policy.

No consensus

Most everyone agrees that network slicing is going to be important for 5G, but there is no industry consensus yet about how to implement it. Deutsche Telekom’s Franz Seiser, Vice President, Core Network and Services, sees this as a problem for operators looking to sell enterprise customers on the benefits.

“We need a common story as an industry to approach the verticals and tell them a consistent story – the basic thing they get with slicing is always the same,” he said, speaking at the Brooklyn 5G Summit in April he said.

GSMA is beginning to work on this, he said, adding that “the idea is to come to a much better story and come to one story as a telecommunications industry, so that we have good lever to get that business on board.”

TM Forum, NGMN and MEF also are addressing these issues. NGMN has published two white papers, one on 5G definition and design and a more recent paper detailing the management requirements for 5G, which focuses mostly on low-level resource and technology requirements. MEF and TM Forum are focusing above on resource-, service- and business-level integration required to support 5G use cases. For more about how standards bodies can work together on 5G, see this article.

“The good thing is we have all the ingredients to address the problem,” says Laurent Leboucher, Vice President Architecture, Enablers and Security, Orange. “The thing we’re missing is the definition of the customer-facing services (CFS), the basic connectivity and access services 5G will provide in the future. Those CFS need to be defined globally and standardized because we will need to interoperate between carriers.”

Assuring 5G slices

A recent TM Forum Catalyst project called 5G service operations real-time service assurance, which was championed by AT&T, BT, Orange, Telecom Italia, Telenor and Vodafone and included MYCOM OSI, Netcracker and TEOCO as participants, focused specifically on how to manage a 5G environment at the service-management level using network slicing and assurance.

The project looked at the efficiency or utilization of the infrastructure as one business driver and at two IoT use cases:

  • an ultra-reliable low-latency communications (uRLLC) smart factory of the future application using robots, sensors, actuators and high-definition cameras connected via a 5G network; and
  • a massive machine-type communications (mMTC) connected vehicle example with a car sending IoT data feeds via a 5G network to an automobile manufacturer to indicate performance.

The idea was to create different types of network slices with completely different characteristics, but using a common infrastructure, and then optimize usage of the resources. For example, if there was no demand, the slices wouldn’t consume resources, or if the uRLLC application needed additional resources to meet its service level agreements (SLAs), then capacity could be increased on the fly, perhaps by ‘stealing’ resources from the mMTC application.

“The challenge is to optimize resource usage,” says Milind Bhagwat, Enterprise Architect, BT. “We don’t want these slices to consume resources when there is no demand, so there has to be constant update from the network in terms of the capacity utilization and quality of service. Based on the data received from the network, optimization has to happen, which means we need to have logic in our orchestration layer to optimize the use of the resources.”

Watch this video to learn more about the 5G service operations project:

What are the challenges

We conducted a survey of 61 telecom professionals from 52 companies for our upcoming Trend Analysis report 5G: Is platform the killer use case?, which will be published at the end of June, asking about platform and 5G challenges including network slicing. More than half of respondents see end-to-end management as big issue.

OSS/BSS of the future

About a third of the CSPs and suppliers we surveyed said availability of OSS/BSS that can accommodate a wide range of innovative business models with multiple parties, using a variety of pricing strategies is a top concern when it comes to slicing.

Systems must be rearchitected for the cloud so that new services can be created almost instantaneously. OSS, for example, need to be based on a modular microservices architecture that follows the principles below. These principles have been identified by TM Forum’s Zero-touch Orchestration, Operations and Management (ZOOM) project (and are explained in greater detail in this white paper):

  • loose coupling of components so that changes to one don’t affect another;
  • exposure of capabilities through standard APIs;
  • policy-driven autonomic support with zero-touch orchestration, network self-healing and self-organization; and
  • metadata-driven and catalog-based with self-declared and well-described components.

How to organize operations?

Our survey also found that 29 percent of CSPs and 26 percent of suppliers are worried about organizational issues when it comes to network slicing. Today telco operations teams typically are arranged vertically around specific product groups, but this likely will not work for platform-based models that use network slicing.

“I think where were headed is that you’ll have an operations responsibility within the service provider’s organization where one group is responsible for service types and another is responsible for efficient use of the infrastructure,” says Dave Milham, Chief Architect, TM Forum. This means CSPs likely will have to organize themselves more horizontally, the way cloud operators do.

Edge-computing resources

About 15 percent of CSPs and 11 percent of suppliers said they are concerned about a lack of edge-computing resources for 5G network slicing. As AT&T’s Kathy Meier-Hellstern, Assistant Vice President of Inventive Science, AT&T Labs, noted in our Quick Insights report Data Analytics & AI: Key to end-to-end management, there is not enough computing power at the edge today to analyze network usage patterns and ensure quality of service.

“One of the challenges with 5G is that we’ve got these very dense access networks with thousands or tens of thousands of locations, as opposed to before where you maybe had hundreds,” she says. “It will be essential to have real-time analytics to support some of our new services like augmented or virtual reality and tactile internet, because they do rely on having analytics at the edge with fairly decent processing and very low latency.”

Milham agrees that solving challenges at the edge is crucial.

“In fixed networks, the core has so much capacity it’s hardly worth worrying about it,” he says. “Where the problems arise is in the access network. You may have one cell running both types of communications where uRLL is on the verge of breaking the SLA agreement. We need to figure out what to do when that happens. The finite resource is the bandwidth in the cell.”

Get some further insight in our new report 5G: Is platform the killer use case? All employees of TM Forum member companies can download it free by registering on our website, and non-members will soon be able to download an executive summary or purchase the full report.



Advertisement:
Share.

About The Author

Managing Editor

Dawn Bushaus began her career in technology journalism in 1989 at Telephony magazine, which means she’s been writing about networking for a quarter century. (She wishes she didn’t have to admit that because it probably gives you a good idea of how old she really is.) In 1996, Dawn joined a team of journalists to start a McGraw-Hill publication called tele.com, and in 2000, she helped a team at Ziff-Davis launch The Net Economy, where she held senior writing and editing positions. Prior to joining TM Forum, she worked as a freelance analyst for Heavy Reading.

Leave A Reply

Back to top