PARTNER FEATURE: As the telecom industry moves towards cloudification and 5G, service providers face the critical challenge of basing quality of experience (QoE) assurance on obsolete methodologies. Legacy traffic monitoring tools (probes) will effectively be blind while migrating telco network functions onto virtual machines. A new approach to achieving visibility into network traffic is therefore required.
This is an inescapable challenge for service providers. Utilising self-organized networks (SON) and network functions virtualization (NFV), 5G must dynamically scale network functions to serve increasing traffic and applications on demand. Different network slices will be created to enable a wide portfolio of services and they will be run in a cloud-based environment. Each slice will run on, and be automatically configured by, multiple virtual machines.
The complexity of this environment should not be underestimated. Multiple, independent, end-to-end networks will be operating over the same infrastructure. In parallel, the networks must ensure maximized spectrum utilization and availability for applications. What’s more, the control plane and user plane separation (CUPS) turns control and user planes into completely independent traffic streams that still need to interact and correlate with each other.
Separating the network from the service means that legacy probes will have no insight into east-west traffic between virtual machines as they spin up multiple networks to support user demands. Traditional separations in network architecture no longer exist, adding even more complexity. This requires that today’s multi-play service providers collect data from a large number of sources including physical and virtual network functions, networking, orchestration and hosting infrastructures.
This telco evolution sees the paradigm shifting from a network-focused industry – in which customers are passive utilizers of the network with minimal control over its attributes – to a user-centric one, where the user controls network functions as well as its dimensions to best serve their own needs.
While it is easy to identify a decrease in quality of service (QoS) through monitoring traditional key performance indicators (KPIs) using passive probes, these are not equipped to drill down into the detail of an experience. New services – such as IoT- present additional challenges: ultra-low latency, enormous volume of devices and unpredictable traffic requirements are just some of the complexity enablers that are overcomplicating QoS assessment.
Traditionally, QoE is evaluated through a set of well-established network service indicators. However, the new user centricity is driving the market to focus on the quality perceived by final users rather than the quality of the network service itself.
What is needed is a means to connect, correlate and analyse data from services, devices, locations and end users. Although performance data is being gathered from these areas and from the network itself, the data, in its rawest form, does not provide insight into QoE. True value is derived by correlating and analysing that data and by constructing a holistic picture of QoE. Companies already doing this can extract even more value from historical data sets by observing trends and using predictive analytics to prevent future issues from occurring.
Holistix, from Empirix, is the analytics platform that enables this capability. Holistix incorporates three distinct types of analytics: descriptive, diagnostic and advanced. Descriptive analytics focuses on what happened, diagnostics analytics explains why it happened and advanced analytics goes to a new level of granularity, answering more complex questions to define the next best action, enabling proactive action plans and assessing the impact of any outage or investments to the company’s return on investment (ROI).
Empirix’s Quality of Experience Index (QoE-I) involves networks, clouds, services, applications, devices and end-users’ performance metrics, all of which come together to provide a complete picture of perceived quality. This is complex in itself but Empirix goes far deeper than simply creating an overall picture of the experience, it moves this capability into the predictive domain. This is achieved through the introduction of predictive analytics capabilities that turn the QoE-I into a predictive index enabling the rapid identification of quality of experience affecting issues.
Predictive QoE-I is based on meaningful key quality indicators (KQIs) collected from a variety of sources. It then uses machine learning algorithms to correlate data and transform it into actionable insights. Key strengths of this solution are that it is service and application agnostic, easily configurable and enables subjective quality assessment methodologies to be applied.
Predictive QoE-I has a broad range of use cases and vertical applications including voice, audio, video, online gaming and IoT.
The value of this approach has been demonstrated in a recent Holistix deployment to support a large service provider’s IoT initiative. Data typically hosted in separate silos, such as network probes, provisioning systems, trouble ticketing, policy management and others, are collected, normalized, and correlated into a single dashboard to provide a consistent assessment of QoE. This allows dispersed operators to assess the performance of a single IoT device, in a specific time range, via a streamlined view, delivering increased operational efficiency.
New insights into the current status of the service or experience can be achieved. For example, the system can know the last time the IoT platform has been accessed, it can validate the provisioning status, and monitor the established service level agreement (SLA).
The important point here is that Holistix is bringing together data from multiple sources, not just Empirix probes. Unlike traditional big data approaches where data is extracted from their original sources, Holistix combines world-class data mediation, correlation and telco expertise to deliver timely, predictive actionable insights.
Combining data from multiple platforms to assess QoE provides enormous value for the business. What’s more, QoE-I extracts a multitude of experience quality information from beyond the network to provide highly granular yet standardized insights into all aspects of service performance.
To learn more about how QoE-I works and how it can be applied to networks of all types as the industry’s move to network virtualization, 5G and exciting new opportunities in IoT, read the Empirix QoE-I whitepaper ‘QoE in the Digital Transformation Era’ here.