After moving my blog to https://garywong.pro, this is my first technical article under the new branding.
Instead of migrating the old post 1:1, I rebuilt the entire SR-MPLS lab from scratch — this time using Containerlab, version control, and documented architecture decisions.

Why I rebuilt this lab

I originally built this topology on EVE-NG back in 2023. It worked, but it also had typical issues:

  • No version control
  • Risk of silent config drift
  • Hard to reproduce environments
  • CLI screenshots instead of structured notes

On Containerlab, things changed:

  • Topology as YAML (Infrastructure as Code)
  • All configs tracked in GitHub
  • Topology image embedded in README
  • Fast teardown / rebuild cycle

This is closer to how real network testing should be done in 2026.

Architecture Summary

The lab focuses on Segment Routing over MPLS using Arista cEOS, with ISIS for the underlay and BGP EVPN for service signaling.

Key design points

  • Single ISIS level-2 domain
  • SRGB: 400000–410000
  • Prefix-SID allocation for all loopbacks
  • Two route-reflectors for BGP signaling
  • EVPN used without VXLAN (pure RFC7432 signaling over MPLS)

Topology image:

Topology overview

Topology YAML

Everything starts from one file:

containerlab deploy -t ceos-sr-mpls.yaml

The YAML defines:

  • Nodes (cEOS images)
  • Management IPs
  • SR-MPLS links
  • Startup configs

The full YAML is here:
https://github.com/svenuscf/arista-ceos-sr-mpls/blob/main/ceos-sr-mpls.yaml

GitHub Repository

Project Repo:
https://github.com/svenuscf/arista-ceos-sr-mpls

Key folders:

  • ceos-sr-mpls.yaml — topology definition
  • configs/ — all router configs
  • arista-ceos-sr-mpls-topo.png — visual topology image
  • readme.md — quick deployment guide

Clone:

git clone https://github.com/svenuscf/arista-ceos-sr-mpls.git  
cd arista-ceos-sr-mpls  
containerlab deploy -t ceos-sr-mpls.yaml

Validation approach (Architect mindset)

This lab is not a config copy-paste article.
I validated the design by answering these questions:

  • Does ISIS correctly advertise prefix-SIDs for every loopback?
  • Is the LFIB populated with the expected next-hops for each remote SR endpoint?
  • Do RR nodes propagate EVPN routes (RT2/RT5) without VXLAN?
  • Does the SRGB remain consistent across all nodes?

When everything aligns, the control plane becomes predictable — and troubleshooting becomes trivial.

Useful show commands

These are the main operational commands I use to verify SR-MPLS behavior on Arista:

show isis segment-routing  
show isis neighbors  
show bgp evpn summary  
show mpls lfib route  
show ip route detail

(Full operational checklist will be a separate post.)

Lessons learned during rebuild

  • Running topology as code is liberating — I can destroy and redeploy the entire lab in <30 seconds
  • Topology changes become Git commits, not memories
  • Using an explicit SRGB range helps avoid silent SID collisions
  • EVPN without VXLAN is still highly relevant for MPLS service design

EVE-NG vs Containerlab (one architect’s perspective)

Feature EVE-NG Containerlab
Config version control screenshots Git
Topology definition UI YAML
Reproducibility Medium 100%
Build speed Slow Fast
Integration Standalone DevOps-ready

Verdict:
I will still use EVE-NG for GUI, but all future design work will start in Containerlab.

What’s next

I plan to expand this series into multiple posts:

  • Part 2 — TI-LFA testing with controlled link failures
  • Part 3 — EVPN Route-Types deep dive
  • Part 4 — SRv6 equivalent lab

Contact

If you’re a network architect, SE, DC/SP engineer, or studying SR-MPLS design — connect with me:

Meta

This is blog post #1 after rebranding from my old platform to Hugo + Containerlab + real Git workflows.
I want my technical content to reflect how I think as an architect, rather than just showing lines of CLI.

Stop writing CLI. Start validating design.