Arista EOS lab 7 of 20
eBGP leaf-spine underlay with ECMP and BFD
Build the routed underlay of a two-spine, two-leaf fabric the way Arista designs recommend: one private AS per leaf, a shared spine AS, peer groups, dynamic BGP peering on the spines, route-map-controlled advertisements, four-way ECMP, BFD and jumbo MTU.
Run this lab
It needs Standard: its 6 nodes are set to 6 vCPU and 9,216 MB in total.
In your workspace: Shared, then Arista EOS guided labs, then 07 - eBGP leaf-spine underlay with ECMP and BFD
- Level
- Intermediate
- Time
- About 80 minutes
- Nodes need
- 6 vCPU, 9,216 MB
- Smallest plan
- Standard
The scenario
Northwind's first production pod is cabled: two spines, two leaves, every leaf connected to every spine with a routed /31 (addresses are already configured). The design standard says eBGP everywhere: the spines share AS 65000 and each leaf gets its own private AS, so a leaf never learns a route back through itself and every path is visible in the AS path. Your job is to build that underlay, prove both spines are used at once, and make failures heal in well under a second.
| Device | Loopback0 (router ID) | AS | Fabric links |
|---|---|---|---|
| spine1 | 10.255.0.1 | 65000 | Et1 10.1.1.0/31 → leaf1 · Et2 10.1.1.2/31 → leaf2 |
| spine2 | 10.255.0.2 | 65000 | Et1 10.1.2.0/31 → leaf1 · Et2 10.1.2.2/31 → leaf2 |
| leaf1 | 10.255.0.11 | 65001 | Et1 10.1.1.1 → spine1 · Et2 10.1.2.1 → spine2 · Vlan10 10.10.10.1/24 |
| leaf2 | 10.255.0.12 | 65002 | Et1 10.1.1.3 → spine1 · Et2 10.1.2.3 → spine2 · Vlan20 10.20.20.1/24 |
| h1 / h2 | – | – | 10.10.10.11/24 via .1 / 10.20.20.11/24 via .1 |
What you’ll practise
- Explain the eBGP underlay AS design (spines share an AS, each leaf has its own)
- Configure BGP with peer groups on leaves and dynamic listen ranges with a peer filter on spines
- Advertise only loopbacks and host subnets with a prefix-list and route-map
- Enable ECMP and verify equal-cost paths through both spines
- Speed up failure detection with BFD and test a spine link failure
- Set jumbo MTU on fabric links and explain why VXLAN needs it
The tasks
- Prepare the links: jumbo MTU
- BGP on the leaves with a peer group
- Dynamic peering on the spines
- Advertise exactly what the fabric needs
- Verify ECMP and end-to-end reachability
- Fast failure detection with BFD
The Lab Tasks pane has the scenario, these tasks, the checks to run, hints and the full solution.
Topology and devices
The network as it opens in EVE-NG, drawn from the lab’s own canvas, and the image each node runs.
Every link, as text
- spine1
Ethernet1to leaf1Ethernet1 - spine1
Ethernet2to leaf2Ethernet1 - spine2
Ethernet1to leaf1Ethernet2 - spine2
Ethernet2to leaf2Ethernet2 - leaf1
Ethernet3to h1eth0 - leaf2
Ethernet3to h2eth0
| Node | Device | Image in Add Node | Set to |
|---|---|---|---|
| spine1 | Arista vEOS | veos-4.33.1.1F | 1 vCPU, 2,048 MB |
| spine2 | Arista vEOS | veos-4.33.1.1F | 1 vCPU, 2,048 MB |
| leaf1 | Arista vEOS | veos-4.33.1.1F | 1 vCPU, 2,048 MB |
| leaf2 | Arista vEOS | veos-4.33.1.1F | 1 vCPU, 2,048 MB |
| h1 | Alpine Linux | clbalpinebase-alpine-base-3-20-3 | 1 vCPU, 512 MB |
| h2 | Alpine Linux | clbalpinebase-alpine-base-3-20-3 | 1 vCPU, 512 MB |
How to open this lab
Sign in to your workspace at
lab.cloudlabbox.com.In File Manager, open Shared, then the folder
Arista EOS guided labs.Open
07 - eBGP leaf-spine underlay with ECMP and BFDand start its nodes, then open Lab Tasks for the scenario and the steps.
The lab is read-only. You can start the nodes and configure the devices, but changes to the topology can’t be saved. To change it, rebuild it in your own folder.
More Arista EOS labs
Related labs in other courses
CloudLabBox is independent of Arista Networks. The lab follows exam topics and isn’t vendor training material; vendor names identify the exams and the images in the workspace.