EVE-NG Split-Approach Network Lab

Project Summary

Field Value

PRJ ID

PRJ-2026-04-eve-ng-lab

Owner

Evan Rosado

Priority

P1

Category

Infrastructure / Lab

Status

Active — Phase 0 (Planning & Documentation)

Purpose

Comprehensive network lab for Cisco portfolio (WLC, ISE, FTD/FMC, ASA, IOS/IOS-XE/XR/NX-OS), multi-vendor (Palo Alto, Fortinet, Arista, Juniper), Linux, and API/DevOps/SecOps workflows.

Approach

Split deployment — lightweight labs on Razer workstation (64GB), heavy labs on kvm-01 (128GB). EVE-NG CE on both as KVM VMs (not bare metal).

Image Access

Active CCO/SmartNet — full Cisco image library available

Resource Allocation

Environment Role RAM Budget What Runs Here

Razer workstation (64GB)

Lightweight labs

~40GB (20GB reserved for desktop)

Multi-vendor peering, R&S, API/DevOps, Linux endpoints

kvm-01 (128GB)

Heavy labs

~50-60GB (after existing VMs)

FMCv, FTDv, ISE, NX-OSv 9000, ASAv, full security stack

Deployment Status

Phase Description Status Notes

0: Planning

Documentation, IP reservation, project structure

🟡 In progress

1: Workstation Deploy

EVE-NG CE as KVM VM on Razer (Arch stays intact)

❌ Not started

16 vCPU, 40GB RAM, nested KVM with host-passthrough

2: kvm-01 Deploy

EVE-NG CE as KVM VM on kvm-01 for heavy topologies

❌ Not started

8 vCPU, 60GB RAM, bridged to br-mgmt

3: Image Library

Cisco + multi-vendor images from CCO and vendor portals

❌ Not started

Lightweight on workstation, full catalog on kvm-01

4: Foundation Topologies

Multi-vendor peering + API lab (workstation, ~20GB)

❌ Not started

IOS-XE, vEOS, vJunos, Palo Alto, FortiGate — all API-enabled

5: API/DevOps Integration

RESTCONF, NETCONF, gNMI, Ansible, pyATS formalized

❌ Not started

Runs parallel with topology building

6: Heavy Topologies

FTD/FMC/ISE security stack, VXLAN/EVPN, MPLS, Wireless

❌ Not started

kvm-01 only — 22-50GB per topology

7: Production Integration

Bridge kvm-01 labs to BIND, ISE, Vault, Wazuh

❌ Not started

VLAN 50 with VyOS firewall restrictions

Assessment

Why EVE-NG (Not GNS3, Not CML)

Platform Strengths Why Not Chosen

EVE-NG CE

Multi-vendor native (Cisco, Palo Alto, Fortinet, Arista, Juniper, Linux). Web UI. Runs QEMU/KVM images directly. No vendor lock-in. Free.

Chosen

GNS3

Good for basic Cisco labs. Client-server model.

Vendor-centric, limited multi-vendor. Client GUI required. Less production-oriented.

Cisco CML

Official Cisco images. DevNet integration.

Cisco-only. Licensed ($199/yr+). No multi-vendor support.

Split Approach Rationale

Problem: No single available host can run the full topology catalog simultaneously.

  • FMCv (28GB) + ISE (16GB) + FTDv (8GB) = 52GB before any other nodes

  • NX-OSv 9000 VXLAN fabric (6 nodes x 8GB) = 48GB

  • Razer workstation has 64GB but needs ~20GB for desktop (Hyprland, Claude Code, browsers)

  • kvm-01 has 128GB but ~63GB already committed to production VMs

Solution: Two EVE-NG instances, each as a KVM VM (not bare metal).

Concern Workstation kvm-01

OS impact

None — EVE-NG runs as VM, Arch/Hyprland untouched

None — EVE-NG is another VM alongside existing fleet

Nested KVM

Yes — ~15-30% overhead on CPU-heavy nodes

Yes — same overhead, but only for heavy nodes that need raw power

When to use

Daily API/DevOps development, multi-vendor comparison, R&S practice

Security stack labs (FMC/FTD/ISE), data center fabric (NX-OS), MPLS, wireless

Network access

NAT or bridged to home network

Bridged to br-mgmt (10.50.1.x), Cloud0 for production integration

Hardware Inventory

Spec Razer Workstation kvm-01

CPU

Intel Core Ultra 9 275HX (24 cores)

Xeon D-2146NT (8C/16T)

RAM

64GB DDR5

128GB ECC DDR4

Storage

2TB Samsung NVMe (188GB free on root)

Production storage (allocate 500GB qcow2)

Nested KVM

VT-x confirmed

VT-x + nested enabled

EVE-NG VM allocation

16 vCPU, 40GB RAM, 200GB disk

8 vCPU, 60GB RAM, 500GB disk

Risk Assessment

Risk Likelihood Impact Mitigation

Nested KVM performance penalty on heavy images (ISE, NX-OSv)

High

Medium

Heavy images on kvm-01 only. Pin to known-good versions. Allocate extra RAM.

Desktop unresponsive when running large topologies on workstation

Medium

Medium

Limit workstation topologies to ~30GB. Shut down unused nodes.

Cisco image licensing enforcement

Medium

High

Active CCO access. Keep SmartNet current.

Lab bridge leaks traffic to production

Low

Critical

VyOS firewall on VLAN 50. Default-deny. No WAN route from lab.

kvm-01 existing VMs degraded by EVE-NG

Medium

High

Monitor with Wazuh. Set cgroups limits on EVE-NG VM. Don’t run heavy topologies during production hours.

Metadata

Field Value

PRJ ID

PRJ-2026-04-eve-ng-lab

Author

Evan Rosado

Created

2026-04-05

Last Updated

2026-04-05

Status

Active — Phase 0 (Planning)

Category

Infrastructure / Lab

Priority

P1

Environments

Razer workstation (lightweight), kvm-01 (heavy)

Edition

EVE-NG Community Edition