Uploaded September 2026 | Updated September 2026, 4 days ago
In this demo, we show how F5 NGINX Gateway Fabric's new ExternalLoadBalancer custom resource bridges the Kubernetes Gateway API with F5 BIG-IP — dynamically, with zero CIS code changes and no new CRDs on the CIS side.
You'll see:
- How F5 CIS acts as the bridge between BIG-IP and your Kubernetes cluster
- Why pod-level tracking creates configuration churn at scale — and how ExternalLoadBalancer solves it
- Automatic Virtual Server provisioning via AS3 and IPAM
- PROXY Protocol preserving full client IP identity end-to-end
- Real-time pod scaling with zero BIG-IP configuration churn
Resources:
- NGINX Gateway Fabric docs: docs.nginx.com/nginx-gateway-fabric/external-loadbalancers/big-ip/quickstart
- F5 CIS IngressLink docs: clouddocs.f5.com/containers/latest
Timestamps :
0:00 Intro — Gateway API evolution of IngressLink
0:15 What is F5 CIS and why it matters
0:35 The problem: ephemeral pod IPs and BIG-IP config churn
0:51 Introducing ExternalLoadBalancer in NGINX Gateway Fabric
1:00 How the architecture works — gatewayLink, IngressLink, multi-cluster
2:00 Demo: cluster walkthrough — Gateway and HTTPRoute resources
2:22 Creating the ExternalLoadBalancer custom resource
3:00 BIG-IP: Virtual Server auto-provisioned via AS3
3:30 Testing the data path — coffee and tea microservices
4:02 PROXY Protocol: client IP preserved end-to-end
4:20 Dynamic scaling demo setup
4:29 Scale up: 2 → 5 replicas, zero BIG-IP churn
4:56 Scale down: connections drain cleanly
5:05 Recap: ExternalLoadBalancer, PROXY Protocol, decoupled architecture
▬▬▬ ABOUT DEVCENTRAL ▬▬▬
We're DevCentral: the how of F5, since 2003. A community of engineers, architects, and IT folks pushing each other forward, one hard-won answer at a time. We dig in, tackle the hard stuff, figure out what actually works, and share what we learn.
Keep going:
Forums, articles, CodeShare: community.f5.com
LinkedIn: linkedin.com/showcase/f5-devcentral
X: https://x.com/devcentral
The DevCentral advocates behind the videos:
Alessandro Fael Garcia → linkedin.com/in/alessfg · github.com/alessfg
Buu Lam → linkedin.com/in/buulam · https://x.com/buulam
Chase Abbott → linkedin.com/in/chaseabbott1
Jason Rahm → linkedin.com/in/jrahm · https://x.com/jasonrahm
Marko Sluga → linkedin.com/in/markosluga
Scott McAllister → linkedin.com/in/stmcallister
In this demo, we show how F5 NGINX Gateway Fabric's new ExternalLoadBalancer custom resource bridges the Kubernetes Gateway API with F5 BIG-IP — dynamically, with zero CIS code changes and no new CRDs on the CIS side.
You'll see:
- How F5 CIS acts as the bridge between BIG-IP and your Kubernetes cluster
- Why pod-level tracking creates configuration churn at scale — and how ExternalLoadBalancer solves it
- Automatic Virtual Server provisioning via AS3 and IPAM
- PROXY Protocol preserving full client IP identity end-to-end
- Real-time pod scaling with zero BIG-IP configuration churn
Resources:
- NGINX Gateway Fabric docs: docs.nginx.com/nginx-gateway-fabric/external-loadbalancers/big-ip/quickstart
- F5 CIS IngressLink docs: clouddocs.f5.com/containers/latest
Timestamps :
0:00 Intro — Gateway API evolution of IngressLink
0:15 What is F5 CIS and why it matters
0:35 The problem: ephemeral pod IPs and BIG-IP config churn
0:51 Introducing ExternalLoadBalancer in NGINX Gateway Fabric
1:00 How the architecture works — gatewayLink, IngressLink, multi-cluster
2:00 Demo: cluster walkthrough — Gateway and HTTPRoute resources
2:22 Creating the ExternalLoadBalancer custom resource
3:00 BIG-IP: Virtual Server auto-provisioned via AS3
3:30 Testing the data path — coffee and tea microservices
4:02 PROXY Protocol: client IP preserved end-to-end
4:20 Dynamic scaling demo setup
4:29 Scale up: 2 → 5 replicas, zero BIG-IP churn
4:56 Scale down: connections drain cleanly
5:05 Recap: ExternalLoadBalancer, PROXY Protocol, decoupled architecture
▬▬▬ ABOUT DEVCENTRAL ▬▬▬
We're DevCentral: the how of F5, since 2003. A community of engineers, architects, and IT folks pushing each other forward, one hard-won answer at a time. We dig in, tackle the hard stuff, figure out what actually works, and share what we learn.
Keep going:
Forums, articles, CodeShare: community.f5.com
LinkedIn: linkedin.com/showcase/f5-devcentral
X: https://x.com/devcentral
The DevCentral advocates behind the videos:
Alessandro Fael Garcia → linkedin.com/in/alessfg · github.com/alessfg
Buu Lam → linkedin.com/in/buulam · https://x.com/buulam
Chase Abbott → linkedin.com/in/chaseabbott1
Jason Rahm → linkedin.com/in/jrahm · https://x.com/jasonrahm
Marko Sluga → linkedin.com/in/markosluga
Scott McAllister → linkedin.com/in/stmcallister




![Agentic Observability: Monitoring AI Traffic with F5 NGINX & OpenTelemetry
AI agents are fundamentally changing infrastructure traffic — and traditional load balancers are flying blind. In this demo, we show how to gain deep, real-time visibility into AI/MCP traffic natively using F5 NGINX and OpenTelemetry — with zero backend code changes.
Using NGINX JavaScript (NJS) at the proxy layer, we parse JSON streams in real time to extract tool names, client identities, and target MCP servers. Combined with OpenTelemetry and the golden signals (latency, throughput, error rates), you get a comprehensive observability layer for agentic AI infrastructure — exportable to any OTel-compatible backend.
🔗 Get the code and try it yourself: [Ihttps://github.com/nginx/nginx-mcp-js/tree/main]
🔗 MCP Blog: [https://blog.nginx.org/blog/introducing-agentic-observability-in-nginx-real-time-mcp-traffic-monitoring]
Timestamps:
0:00 – Introduction & the AI traffic problem
0:22 – Why AI agents are unpredictable (fan-out behavior)
0:46 – Three critical blind spots: delivery, UX, and observability
1:36 – The NGINX solution: native Layer 7 MCP inspection
2:07 – How it works: NJS parsing + OpenTelemetry golden signals
2:51 – Demo walkthrough: installation & NGINX config
3:12 – Extracting MCP metadata and mapping to OTel spans
3:53 – Running the load generator to simulate AI traffic
4:00 – Grafana dashboard overview
4:19 – Troubleshooting: isolating slow tools
4:46 – Identifying rogue AI clients by identity
5:05 – Monitoring backend MCP server health
5:32 – Wrap-up and next steps
Notes:
- Contributed by: Akash Ananthanarayanan
- Related Article: None
⬇️⬇️ JOIN THE COMMUNITY! ⬇️⬇️⬇️
DevCentral is an online community of technical peers dedicated to learning, exchanging ideas, and solving problems—together.
Find all our platform links ⬇️ and follow our Community Evangelists! 👋
➡️ DEVCENTRAL: https://community.f5.com
➡️ YOUTUBE: https://youtube.com/devcentral
➡️ LINKEDIN: https://www.linkedin.com/showcase/f5-devcentral/
➡️ X: https://x.com/devcentral
Your Community Evangelists:
👋 Jason Rahm: https://www.linkedin.com/in/jrahm/ | https://x.com/jasonrahm
👋 Buu Lam: https://www.linkedin.com/in/buulam/ | https://x.com/buulam
👋 Aubrey King: https://www.linkedin.com/in/aubreyking | https://x.com/aubreykingf5
👋 Chase Abbott: https://www.linkedin.com/in/chaseabbott1 Agentic Observability: Monitoring AI Traffic with F5 NGINX & OpenTelemetry](https://i.ytimg.com/vi/PydaFJirjj0/mqdefault.jpg)





