Deep Network Traffic Observability with Pktvisor and Prometheus
summary
Open source pktvisor is a C++ edge agent that captures packet and application-layer fields (DNS example) and uses streaming algorithms to produce top-k heavy hitters, cardinality estimates, quantiles and histograms in 1-minute buckets. It exposes metrics natively for Prometheus scrape or remote_write, runs as a container or static binary, filters at BPF/kernel level (example: port 53), and supports concurrent policies per interface. A complementary orb control plane uses MQTT and a Helm chart for fleet policy orchestration and multi-tenant agent management.
AI-written summary · about
chapters (15)
- 0:00Intro
- 1:08What is deep network observability
- 6:15What is Pktvisor
- 13:47Pack Advantage
- 14:27Prometheus Ecosystem
- 14:54Pktvisor
- 15:20Sample dashboards
- 15:36Viewing metrics
- 15:56Premade dashboard
- 16:19Orb Project
- 16:58Orb Platform
- 17:51Orb Agent
- 19:45Orb User Interface
- 20:25Summary
- 20:44Questions
related talks
35mWho Dropped the Packet? Solving K8s Network Mysteries in Real-Time
31mZero instrumentation observability with eBPF
20mNetwork Monitor: A Tale of ACKnowledging an Observability Gap
34mNetwork Observability with eBPF and OpenTelemetry
40mBuilding a Scalable Observability Platform for Edge Compute Nodes
31mA High-Schooler’s Guide to Kubernetes Network Observability
31mFrom Kernel To Cloud: Building a Complete Open Source Observ...
26mNetwork monitoring with FOSS