Platform Telemetry: Fundamental for Automating Power Efficiency, Network Observability

Platform Telemetry: Fundamental for

Automating Power Efficiency, Network

Observability 

It’s just a matter of time before mass adoption of cloud native deployments in 5G networks because the architecture will deliver the scale and flexibility required to cost-effectively manage data growth and generate monetization opportunities.  

According to an Intel-sponsored IDC survey, one of the biggest operational challenges for network infrastructure managers and executives is the lack of visibility across the infrastructure and complete network stack, which can negatively affect quality of service and customer satisfaction. This at a time when communication service providers (CoSPs) are focusing on energy reduction and renewable energy sources to meet sustainability goals and cut operational expenses. 

Telemetry data generated by processors provides the foundational insights for CoSPs to effectively measure and automate the management of infrastructure utilization, power consumption, fault detection and resolution, and performance in cloud native networks.  Yet many CoSPs still do not leverage telemetry to its full potential.  

Let’s examine how Intel® Platform Telemetry makes it possible for CoSPs to drive energy efficiency and deploy full-stack visibility and service assurance solutions across the 5G network infrastructure.  

Measure & Manage Power: Reduce Energy, Carbon & Costs 

The first step toward improving energy efficiency is to determine and optimize power consumption. Telemetry enables this measurement and subsequent actions that can reduce power consumption, carbon emissions, and operational expenses.  

Server power consumption is a significant opportunity for energy savings. Network operators can leverage telemetry to identify idle, underutilized, or unallocated infrastructure and use platform optimizations to shift to low power mode. A recent proof of concept running AI software from QiO on 3rd Gen Intel® Xeon® processors achieved up to 52 percent power savings when idle, and up to 24 percent with a representative load, on a given CPU.  

Another unique challenge for network operators is, historically, CPUs have drawn the same amount of power regardless of fluctuations in data traffic pattern. Until recently there has not been a way to match real-time traffic with CPU utilization and power consumption.  

Launched at MWC Barcelona 2023, the new Intel® Infrastructure Power Manager for 5G Core enables CPU power savings by dynamically matching CPU performance and energy consumption to traffic. The new software simplifies access to key Intel® Xeon® capabilities, such as power telemetry, granular power control states, and low-latency frequency change.  

Because CoSPs cannot trade power-efficiency for performance, the software reduces power by an average 30 percent1 while maintaining performance and supporting stringent throughput and service level agreement (SLA) requirements. The new solution is already winning praise from Nokia, NEC, Casa Systems, and SK Telecom. 

Telemetry Feeds Full-Stack Observability and Assurance  

 Almost 70 percent2 of CoSPs admit that fault detection and correlation in cloud-based 5G infrastructure is more difficult than in LTE according to a 2022 Heavy Reading study.  Without infrastructure visibility, how do you ensure compliance with internal KPIs and customer SLAs?  

Consider the case of low-latency connectivity to IoT robots in a manufacturing facility. What happens if the network goes down? How long will it take to identify and repair the root cause? What does that downtime cost in terms of penalties, lost productivity, and reputational damage?  

What if you could tap the right data at the right time and apply it toward a specific outcome and KPI, like identifying, correlating, and resolving network faults before they impact SLAs?  

Intel’s collaboration with EXFO serves this exact purpose - to close the visibility gap between cloud-native infrastructure, the network, and service layers. The EXFO full-stack assurance solution consumes rich Intel Platform Telemetry Insights with Open Telemetry; enables CoSPs to assess the operational state of the underpinning cloud-native infrastructure from the chip-level through the network stack; and isolates and correlates faults across all network domains to speed up time-to-resolution and eliminate domain-specific fault denial. 

Telemetry Goes Primetime with Intel and Open-Source Ecosystems 

Technology and IT ecosystems increasingly recognize the value of telemetry data for a wide range of high-value applications. It speaks volumes that Open Telemetry is the most active project within the Cloud Native Computing Forum behind only Kubernetes. In its “Hype Cycle for Emerging Technologies, 2022,” Gartner estimates that both open telemetry and cloud sustainability will reach the productivity plateau in two to five years. 

Intel® Platform Telemetry Insights provide meaningful and actionable data across single or multiple servers, including the Power Insights Kit with open-source tools to collect, manage and visualize power usage and calculate opportunities for energy savings. Telemetry can also enable the calculation of carbon reduction opportunities, which requires data on the carbon intensity of each energy source. Because Intel architecture powers the majority of NFV and cloud-native deployments across the wireless core, RAN, and edge and is also prevalent across private and public clouds around the world, this telemetry data and its insights are available in each deployment. Intel fine-grained telemetry is made available via open-source projects Collectd and Telegraf.  

Our deep expertise in telecommunications is built into processor optimizations and applications. For the past decade, Intel has operated and evolved our in-house network stack. Today, this commercial-grade, 3GPP-standards-compliant, cloud-native 5GC stack runs control plane and user plane network functions in containers, which informs optimizations, reference architectures, open-source contributions, and future silicon. 

 We will continue to add value by applying telemetry data to address emerging use cases and applications, including service assurance platforms, alerts, orchestration, automation, and more.  

Tap Intel Platform Telemetry Insights  

Platform telemetry data is foundational to enable CoSPs with full-stack visibility, measurement and management of power consumption, infrastructure utilization, and performance.  Network engineers, architects and developers can access Intel Platform Telemetry Insights and its broad array of data, reports, and tools to get their arms around their current situation. These resources are available under NDA by visiting the Intel Developer Zone. 

If you would like to take platform telemetry to the next level as part of a comprehensive power management and service assurance strategy, we are here to engage as your partner. 

1 Performance varies by use, configuration, and other factors. See [N35] at www.Intel.com/PerformanceIndex 

2 “5G Troubleshooting and Fault Detection Study, North America and Western Europe 1Q22 Survey Analysis,” Heavy Reading, March 2022