Network operators need to quickly find and resolve latency issues in their 5G networks in order to unlock new revenue streams. Unfortunately, traditional approaches to service assurance often lack the context necessary to analyze and rectify network performance effectively. While accurate network and service inventory data can provide this context, it has been hard to achieve an up-to-date, complete picture of the network. As a result, many network operators lack the ability to visualize and optimize the network service path based on latency, so they can’t achieve the latency targets that would enable them to deliver real-time services and offer premium SLA-backed offerings.
The need for low-latency communications
5G promises new revenue from real-time services ranging from high-definition mobile gaming to mission-critical enterprise applications such as autonomous guided vehicles and video analytics. Each of these is enabled by an ultra-reliable, low-latency communications (URLLC) network, a key aspect of the 5G vision, as well as other 5G capabilities like network slicing. The performance of the transport network underpins this vision, so it’s critical to understand and eliminate sources of latency, such as network segments and network functions.
Each network element contributes to the overall latency of a given service path. As a result, it’s becoming more and more crucial to detect the latency of network segments in real time, before they impact customer experience and premium services.
The cost of latency
Latency has a real-world impact on the delivery of real-time services as well as normal mobile usage. It negatively impacts the perception of network performance, which affects customer satisfaction and ultimately leads to churn. It creates delays in deploying new services—backed by stringent SLAs—since it isn’t clear how the network will perform. Drive tests and remote onsite testing, both of which are OPEX-intensive, are required to validate the network on a recurring basis.
There is an opportunity to optimize the network for latency, from radio unit to mobile core and cloud. In mapping out the topology of the network and the services that it supports, there is an adjacent opportunity for better resource utilization, improved provisioning, and improved customer satisfaction through “first time right” deployments.
Optimizing the network: reducing latency
Leading telecom service providers have set aggressive latency goals such as the “10-millisecond network.” Achieving such goals is another matter entirely. In practice, optimizing the transport network requires an inclusive, end-to-end assurance strategy. All groups—RAN, core, transport network, traffic engineering, network planning, cloud infrastructure—must do their part to shave precious millisecond delays out of each domain, while also identifying and resolving bottlenecks where they occur.
This endeavor calls for a unified view of network performance across all network segments and domains, end-to-end. To do this, ubiquitous, segmented performance monitoring is needed across physical and virtualized infrastructure, from radio unit to mobile core and cloud.
Putting the network in context
Network performance metrics need to be efficiently analyzed in context to provide real-time detection of latency variations within dynamic infrastructure.
However, network operators frequently lack the context to understand how data traverses the network in practice. Inventory and topology data are often stored in siloed, incompatible systems, making it difficult to build a network and services map and keep it updated as the network evolves. Even more challenging is the move to dynamic infrastructure, which sees software-defined network functions constantly instantiated and torn down as network demand ebbs and flows.
This means accurate segment-by-segment performance data is not readily available, making it difficult to make informed decisions about improving latency on the network, let alone offering and achieving stringent performance SLAs.
Even with context, passive monitoring is not enough to deliver a full picture of the performance of the network and the services it supports. Synthetic testing can simulate the real-world experience for subscribers across a range of network and over-the-top (OTT) services. Virtual drive testing, made possible by leveraging real-world RAN performance data, enhances the picture further and can be used to prototype the performance of new services before they launch.
Finally, streaming analytics can reveal anomalies, bottlenecks and their root causes, and prescribe actionable steps to permanently resolve issues and systematically optimize latency.
Latency and the EXFO adaptive service assurance platform
EXFO’s adaptive service assurance platform helps network operators resolve network latency challenges with a combination of dynamic network topology and inventory, cloud-native active test agents to monitor network quality and user experience, source-agnostic data ingestion—including all major RAN equipment suppliers—and multi-domain, multi-layer analytics. As a result, it provides full visibility into (and diagnostics for) complex transport networks and the services and user experience that they support. In short, this EXFO platform enables network operators to efficiently reduce latency and operate 5G-ready transport networks.