Like the introduction of most new services, 5G came with a great deal of hype. Ultra-reliable low-latency communications (URLLC), massive machine-type communications (mMTC) and enhanced mobile broadband (eMBB) promised to change the way people, and perhaps more importantly, machines interact with each other. Service agility, massively scalable networks, ‘any service, anywhere, anytime’, and the holy grail—zero-touch automation—would finally become a reality.
And, while great strides have been made in achieving these promises, like all over-hyped solutions, reality tends to lag somewhat. As the old saying goes, the devil is in the details.
But the real question, after all the money spent in building out a 5G network and investing in automation, is — will customers be willing to pay more for new 5G enabled applications and services?
If history has taught us anything, the answer is yes—but only if they deliver better value.
Multi-Access Edge Computing – Benefits and Challenges
Multi-access edge computing (MEC), also referred to simply as edge computing, brings compute and network functions closer to the end-user. MEC can help deliver a better user experience for low-latency services such as cloud gaming with AR or VR, first-responder video or connected car applications by shortening the distance—and hence lowering the latency—between the user and the real-time application.
But, of course, there are many different types of ‘edge’—hyperscaler edge, core network edge, RAN edge and customer-premises edge, to name a few. The challenge for MEC, as it moves further away from the central office or wire center, is that rack space, power and quite likely, physical access can be quite limited—which creates a new challenge for the mobile service provider (MSP). Unlike the core data center, where power, CPU, GPU, memory and storage can be assumed to be unlimited, the MEC does not have this luxury. Consequently, for applications demanding ‘real-time’ behavior, the MSP will likely need to place the latency-sensitive microservices of the application in the MEC, while the other non-latency-sensitive microservices are deployed in a larger RAN or core location.
It is this distributed architecture that is both a benefit and a challenge to the MSP. While it’s essential for delivering the low-latency applications customers will be willing to pay more for, the distributed nature of the application’s microservices complicates the management of the customer’s quality of experience (QoE), and consequently the success of the applications.
Unlike previous generations of mobile services, 5G is expected to deliver true business-class services. In fact, a recent study suggests that up to 65% of 5G services will have some form of SLA and as much as 50% of the MSP revenue will come from these services. Failure to delight the customer is no longer an option. To manage 5G, and especially MEC enabled 5G services demands a different way of looking at assurance.
Simple connectivity assurance is no longer sufficient. In this new reality, QoE is everything. The MSP needs an assurance solution that delivers visibility from the core-to-edge and infrastructure-to-application. Further, the complexity of this network makes automation a must—and that demands an assurance solution that can work across the traditional ‘siloed operations’ of MSP to the right data (root cause analysis), at the right time (real-time response) and in context (highest ‘business critical’ needs first). The right data, at the right time, and in context is the key to the ‘zero-touch’ holy grail.
As a ‘sign-of-the-times’, an emerging requirement for MEC revolves around power consumption.
Of course, as mentioned, there’s the issue of limited power availability at MEC locations and the need to ensure solutions, especially in multi-tenant MEC locations, stay within their respective budgets. However, the idea of ‘Green initiatives’ is also becoming a concern, both from the regulatory side, but perhaps more importantly, from the Brand image side. Many end-users will start considering such initiatives when selecting their 5G provider. Having detailed visibility of power usage in the MEC is essential to long-term success.
5G Is a Complex Ecosystem
Combining the benefits of MEC with 5G’s high bandwidth and reliability opens the door to new services and applications that require close coordination between cloud hosting and the mobile network. This coordination includes handoffs between cell sites (roaming), Edge Cloud platform locations (various MEC types), distributed network functions and microservices, and application hosting services. With so many moving parts, delivering a seamless user experience certainly isn’t easy!
To begin with, this ecosystem will be massively scalable and highly dynamic—and it’s accepted that the complexity of managing it effectively is well beyond the capabilities of humans and traditional telecom management systems. It’s also understood that if the MSP is to truly automate service quality management end-to-end, they will need to find a way to break through the traditional siloes which stand in the way of root cause visibility to rapidly deliver actionable insight to the network and service orchestrators. To the MSP, this is an entirely new paradigm for operations, but one that must be embraced if they are to succeed.
Underscoring this complexity is the fact that much of the MEC infrastructure involved will be hosted by hyperscalers in Amazon, Microsoft and Google Clouds. Regardless of who owns the contract, enterprise end customers will expect an SLA on application performance. For service providers, this means managing their part of the SLA with real-time orchestration across many domains: RAN, transport, network core, application and network function hosting locations.
A New Fault Domain to Consider
The 5G network, along with the MEC, will be a Cloud-native, microservices based implementation. A key challenge for service assurance is the decoupling of the network and service topologies from the cloud infrastructure. Despite the hype around cloud reliability and the ability of orchestrators to find alternative routes around infrastructure failures, the fact remains that some issues do still go undetected. And in the resource-constrained MEC, even if an infrastructure issue is detected, there may not be an ‘alternative path’.
For the MSP, simply being able to determine the fault domain —infrastructure, application, 5G network, RAN, Core, someone else’s network—and send the right technician to address it within the SLA time constraints, with the right tools can make the difference between violating an SLA and possibly losing a customer or not.
There Is a Light at the End of the Tunnel
The challenges facing the MSP in deploying 5G standalone and MEC are real, but not insurmountable. All hype aside, the industry is making great strides in advancing automation to handle the complexity of these highly agile, massively scalable networks.
And, advances in software-based, containerized microservices for assurance and visibility are being made, almost daily. And that’s good, because for the MSP that’s facing declining or flat average revenue per user (ARPU) and non-trivial, increasing CAPEX costs to build out this 5G network, reducing OPEX while delivering a better experience and new services is essential.
The 5G network will be a data-driven network. In fact, some large Tier 1 provider expect the network will generate upwards of 40 petabytes of data per hour. Traditional assurance solutions cannot handle this volume of data, but more critically, operators cannot afford to transport, store and analyze this amount of data. Assurance systems need to get smarter in how they generate metrics.
Most of the key performance indicator (KPI) data generated simply indicates that the network is running fine—zero incremental value for customer experience management. However, when an issue is detected, the assurance system needs to automatically scale up monitoring and analytics to isolate the issue, correlate it with other issues, identify the root cause and pass that along to be dealt with (either automatically by an orchestrator or raise a trouble ticket).
After confirming the issue has been resolved, the assurance system can scale back to its original level of monitoring.
Developing an automated assurance solution like this is becoming more and more straightforward. Leveraging existing big data techniques, advances in machine learning and artificial intelligence, open-source community reference models and open APIs is encouraging innovation and specialization for telecom operating models.
Mobile edge computing may challenge the MSP when it comes to managing the customers end-to-end QoE, but the industry has moved past the hype and is approaching the challenge with eyes wide open. The promise of 5G may not be upon us yet, but there is a light at the end of the tunnel.
[To share your insights with us, please write to firstname.lastname@example.org]