Cumulus VX leaf and spine switches in EVE-NG
Cumulus VX, from 2.5 to 5.12, is in Add Node in your CloudLabBox workspace. Build Linux-based leaf and spine fabrics with BGP, EVPN and VXLAN, one small node per switch.
Choose a planCheck it in the free demo
- Vendor
- NVIDIA
- In Add Node
- 22 images
- Template
cumulus- Newest
cumulus-vx-5.12.0
Installed versions
Every Cumulus VX name in the Add Node image list, exactly as you’ll pick it.
Started in our checksKnown problemEach links to the tested tip below.
Template cumulus
Cumulus VX
cumulus-vx-5.12.0cumulus-vx-5.9.1Started in our checkscumulus-vx-5.8.0cumulus-vx-5.7.0cumulus-vx-5.6.0cumulus-vx-5.5.1cumulus-vx-5.5.0cumulus-vx-5.4.0cumulus-vx-5.3.1Known problemcumulus-vx-5.2.1Known problemcumulus-vx-5.1.0cumulus-vx-5.0.1
Show 10 older
cumulus-vx-5.0.0cumulus-vx-4.4.1cumulus-vx-4.4.0cumulus-vx-4.3.0cumulus-vx-4.2.1cumulus-vx-4.2.0cumulus-vx-4.1.1cumulus-vx-4.1.0cumulus-vx-4.0.0cumulus-vx-2.5.3
Add it to a lab
In Add Node, choose the template, then the image. The template fills in these settings; change them before you start the node.
| Image | Template | CPU | RAM | Interfaces | Console |
|---|---|---|---|---|---|
| Cumulus VX | cumulus | 1 | 1024 MB | 8 interfaces: eth0, then swp1 to swp7 | telnet |
EVE-NG admits a node by the CPU and RAM set on it, so a running node counts against your plan at these values. This template starts nodes at 1 vCPU and 1,024 MB, which fits the free demo’s 6 vCPU and 8 GB, so you can start one there and open its console.
A leaf and spine fabric with BGP unnumbered
Wire two spines to two leaves on their swp ports, bring up BGP unnumbered for the underlay, then add EVPN and VXLAN so two Tiny Core hosts share a VLAN across the leaves.
Each Cumulus VX node starts at 1 vCPU and 1024 MB, so a whole fabric stays small.
- 4×Cumulus VX
cumulus1 vCPU, 1024 MB each - 2×Tiny Core Linux
clbtinycore1 vCPU, 512 MB each
Added with these defaults, the lab uses 6 vCPU and 5 GB of RAM. The smallest plan that fits is Lite, with 6 vCPU and 8 GB: £3 a day, or £0.50 an hour On-Demand.
Licences are yours to bring
CloudLabBox doesn't provide NVIDIA licences or support. Cumulus VX is NVIDIA's virtual edition for testing, so check NVIDIA's terms before you use it for anything beyond practice.
The images run only inside your CloudLabBox workspace; they aren’t available to download.
Guided labs that use Cumulus VX
Ready-made labs in the Shared folder of every lab workspace, each with a scenario, tasks, checks and the full solution. Here are 8 of the 12 that use it, from 1 course, counted on 2 November 2026.
Questions about Cumulus VX
What is NVIDIA Cumulus VX?
Cumulus VX is the virtual edition of NVIDIA Cumulus Linux, a Linux-based network operating system for data centre switches: BGP, EVPN and VXLAN, configured with NVUE or standard Linux tools.
Is Cumulus VX the same as Cumulus Linux?
It's the virtual edition of the same network operating system, so the commands you learn carry over to switches that run Cumulus Linux.
Which Cumulus VX releases use NVUE?
NVUE, NVIDIA's object-based CLI, is the default from Cumulus Linux 5.0, so pick a 5.x image such as cumulus-vx-5.9.1 for NVUE labs. The 4.x images use NCLU by default.
Cumulus VX: which image should I choose, and which port is management?
cumulus-vx-5.9.1 reached its login prompt. Avoid the installed 5.2.1 and 5.3.1 images until their truncated disks are replaced. In the Cumulus profile, eth0 is management and data ports start at swp1.
Which Cumulus VX images are installed?
22 images, all listed on this page exactly as Add Node shows them. The newest is cumulus-vx-5.12.0.
How much CPU and RAM does a Cumulus VX node use?
When you add one, the cumulus template adds 1 vCPU and 1024 MB of RAM. You can change both in Add Node, and your plan counts what each running node is set to.
More images
CloudLabBox is independent of the vendors named on this page. Their names identify the images in the workspace’s Add Node list.