Skip to content
All work
Cloud Infrastructure & SRE

Multi-VLAN Network Segmentation & Dual-WAN Failover

A segmented home/lab network built to enterprise conventions — VLAN-isolated trust tiers with stateful inter-VLAN firewall rules, automated ISP failover, and DNS-level ad/threat filtering — rather than one flat trusted subnet.

Project Brief

Role
Network architecture and implementation owner
Scope
A segmented home/lab network with 4 VLAN tiers, dual-WAN failover, and headless diagnostic tooling against the platform's own management API
Timeline
Built and iterated in 2026
Result
A network where a compromised or misbehaving IoT device can't reach anything else on the network, internet connectivity survives a primary-ISP outage automatically, and every configuration change is auditable after the fact

Evidence Included

  • Directional isolation, not a blanket block

    Inter-VLAN firewall rules are written to allow legitimate control traffic in one direction while denying lateral movement in the other — verified by testing both directions, not assumed from the rule set alone.

  • Multi-target failover health checks

    WAN failover is driven by multiple independent ICMP and DNS probe targets rather than a single check, avoiding false failovers from one unreliable probe target.

  • Auditable configuration history

    Every VLAN, firewall, and firmware change is logged with a timestamp via a custom audit script, making post-change troubleshooting and change review straightforward.

4 (trusted/lab/IoT/unfiltered)

VLAN tiers

2 (fiber primary + cellular failover)

WAN paths

Multi-target ICMP + DNS

Failover health checks

The Problem

A single flat network puts every device — trusted workstations, IoT devices of unknown provenance, and lab/test equipment — on equal footing: any compromised or misbehaving device can reach anything else on the network. There was also a single upstream internet connection, so any ISP outage took every service offline with no failover path.

My Approach

Multi-Tier VLAN Segmentation

  • Designed a 4-tier VLAN architecture separating trusted general-purpose devices, a lab/test tier for experimentation, IoT devices, and an explicitly unfiltered tier for devices that need unrestricted internet access without exposure to the rest of the network.
  • Wrote stateful inter-VLAN firewall rules that default-deny IoT-to-private-network traffic — an IoT device can reach the internet, but not other VLANs — while still allowing traffic in the direction actually needed for control (e.g., a phone on the trusted VLAN reaching an IoT device).
  • Locked down the gateway's own administrative interfaces (SSH, HTTP management) to be unreachable from any VLAN except the trusted management tier, closing off the router itself as a lateral-movement target.
  • Enabled 802.11r fast BSS transition on the wireless tier so trusted-network clients roam between access points without a re-authentication delay, and layered DNS-over-HTTPS at the gateway so DNS queries aren't visible or spoofable in plaintext on the wire.

Automated Dual-WAN Failover

  • Configured a secondary, independent internet uplink (a cellular backup connection) alongside the primary fiber connection, with policy-based dual-WAN routing on the gateway.
  • Set up multi-target health monitoring (ICMP and DNS probes against multiple independent targets, not a single point of failure in the health check itself) so the gateway detects a real primary-WAN outage and fails over automatically, rather than relying on a single, possibly-unreliable probe target.

Headless Diagnostic & Audit Tooling

  • Built a small, dependency-light Python toolset that talks directly to the network controller's REST API for live diagnostics: real-time port/PoE telemetry, switch-port-to-device mapping, and wireless RF utilization scans, without needing the web dashboard open.
  • Added an automated post-change audit log so every configuration change (VLAN edits, firewall rule changes, firmware updates) leaves a timestamped record, making it possible to reconstruct what changed and when after the fact.

Key Engineering Challenges Solved

01

Getting isolation direction right

A naive "block everything between VLANs" rule also blocks the legitimate control traffic a user actually needs (e.g., reaching a smart device from a phone). Solved by writing directional, stateful rules — deny lateral IoT-initiated traffic, but allow trusted-initiated traffic into the IoT tier for control — rather than a blanket bidirectional block.

02

Failover health-check reliability

A single health-check target can itself go down without the primary WAN actually being down, causing a false failover. Solved with multiple independent monitoring targets across both ICMP and DNS, so a real outage requires multiple correlated signals, not one flaky probe.

Stack

Networking Hardware

Managed PoE switchMultiple wireless access pointsCloud-managed gateway/router

Segmentation & Routing

802.1Q VLAN taggingStateful inter-VLAN firewall rulesPolicy-based dual-WAN routing

Wireless & DNS

802.11r fast roamingDNS-over-HTTPS

Tooling

Python (stdlib-only REST API client)Controller REST/v2 API

Skills Demonstrated

  • Network segmentation and zero-trust-adjacent design at a home/lab scale using the same directional-isolation principles as enterprise networks
  • Failover architecture with multi-target health-check design to avoid false-positive failovers
  • Headless automation against a vendor's REST API when the web dashboard isn't sufficient for repeatable diagnostics
  • Change-auditing discipline applied even in a low-stakes environment, as good practice rather than a compliance requirement