Verizon teams up with Pixie to deliver full observability for the network edge
Verizon 5G Edge with AWS Wavelength delivers increasingly mobile and immersive experiences to consumers by moving compute to the edge. Verizon is heavily invested in 5G networks, the services that run on the network, and infrastructure partners like AWS.
Challenge: Identifying the closest Wavelength Zone to a device at any given point
Verizon 5G Edge with AWS Wavelength natively supports Elastic Kubernetes Service (EKS) with Amazon at the edge, such that developers can extend new or existing clusters to any one of the 10+ Wavelength Zones nationwide. The more Zones, the more ways that phones and other devices have to reach optimal performance.
But having multiple zones also presents an observability challenge: If you have pods somewhere in an Availability Zone and you have pods in a Wavelength Zone, what is the latency difference between those pods? Moreover, how might metrics and logs downstream differentiate between edge and non-edge, which may exhibit different network characteristics?
Solution: Supporting observability needs at the network edge with Pixie
Pixie provides automatic observability into the performance of the Kubernetes applications. As noted in its documentation, Pixie is an open source observability tool for Kubernetes applications. With Pixie, there’s no need for manual instrumentation; instead Pixie uses Extended Berkeley Packet Filter (eBPF) to capture telemetry data automatically. Developers can view the high-level state of their clusters and also drill-down into more detailed views using Pixie.
“Imagine a scenario where you notice that gigabytes on gigabytes of data are being exchanged from the edge back to the parent region. You could actually use these insights to re-architect your edge application and introduce efficiencies in your end-to-end workflows,” says Robert Belson, Corporate Strategy at Verizon. “In this way, Pixie on Verizon 5G Edge can be used not only as an observability tool, but also as an architectural asset.”
Out-of-the-box instrumentation: On the path toward full automation
Verizon and New Relic collaborated on a Terraform module where developers simply needed to provide their Pixie deploy key, Pixie API key, and EKS cluster name. Everything else was built-in. With the EKS cluster, developers could select which Wavelength Zones they wanted their node groups to live in. From there, they could move quickly to auto-instrumentation.
Additionally, Pixie has made network flow monitoring with the Pixie query language (PxL scripts) easier. PxL scripts are flexible, making it easier for Verizon developers to create, customize, and adapt new views.
Performance monitoring is optimized with New Relic
By leveraging New Relic on Verizon 5G Edge, developers can leverage a single dashboard across edge and non-edge workloads, leading to a more holistic visualization of performance and availability. Additionally, the Verizon team has been working to make the Pixie deployment even easier and more automated. Developers can automate the agent install, alerts setup, and dashboard configuration, all with a single line of code.
Looking forward, performance monitoring capabilities with New Relic can be deeply integrated with the Verizon edge discovery service, an API which determines the optimal edge workload for a given mobile client.
“This is how we start to think about network intelligence in the context of Verizon 5G Edge,” says Robert. “It's no longer just about application metrics, CPU, or memory utilization. It's about how network intelligence can deliver unprecedented workload orchestration capabilities.”
Conclusion: Transforming how the world connects with Pixie and New Relic
Today’s customers want more immersive, hassle-free mobile experiences. With Pixie and New Relic, Verizon can deliver on this demand more easily. Pixie auto-instrumentation makes it quick to implement Verizon solutions, while New Relic features, like alerts, provide enhanced observability into Verizon software.
Essentially, the combined capabilities of Verizon and New Relic can be viewed as a guide for building best practices in deploying application performance monitoring (APM) to the edge. They have also created a reference architecture to guide future developers to the right metrics and best ways to configure dashboards and alerts. Together, Verizon, Pixie, and New Relic are transforming how people, businesses and things connect with each other—and how developers innovate on the edge.
- Headquarters: New York
- Fortune rank: 20
- Retail locations: Nearly 1,500
- Fortune 500 customers served: 99%
- Countries served: 150+
- Environment: AWS
- Team overview: Developer Relations for Verizon 5G Edge
- Service used in automation template: Pixie
- In a multi-edge deployment, determine which Wavelength Zone is the most optimal at all times.
- Prevent degradation in performance and poor device and user experiences that Verizon’s API, Edge Discovery Service (EDS) might not surface.
- Leverage telemetry data gathered from Pixie on K8 and application services infrastructure as a part of the routing algorithm to determine the optimal network path of a MEC device (camera).
- Use New Relic to detect connection instability of a MEC device and trigger a call (via webhook) to EDS so that subsequent decisions can be made to route the device to the next best performing network path.