← Back to portfolio
Flagship Infrastructure Project 19.5 / 20

Secure Virtualised Micro Data Centre for Odoo SaaS

A production-grade micro data centre built from the ground up on Proxmox — orchestrating a multi-tenant Odoo SaaS platform across Docker Swarm, pfSense VLAN segmentation, HAProxy high availability, and GitLab CI/CD. Validated under live failure simulation with zero service disruption.

8–12VMs managed
10+VLANs secured
5-nodeDocker Swarm
0Downtime in failover tests
~60%Deploy effort cut by CI/CD
Proxmox VE Docker Swarm pfSense HAProxy PostgreSQL Redis GitLab CI/CD Prometheus Grafana Odoo SaaS NFS
View Source Code Watch Demo
Live Demo

End-to-end demonstration

Demonstration

What you're seeing

The demo puts the platform under real stress — a node is deliberately killed mid-operation, and the infrastructure responds automatically. No manual intervention. No downtime.

  • 1 Node failure simulated — a cluster node is interrupted while Odoo is serving traffic
  • 2 Automatic rescheduling — Docker Swarm detects the failure and reschedules affected containers to healthy nodes
  • 3 Odoo stays reachable — HAProxy continues routing traffic throughout the disruption
  • 4 Redis failover triggered — a replica takes over the cache layer without data loss
  • 5 Live monitoring throughout — Grafana dashboards show cluster state, request rate, and recovery in real time
Back to Home
Architecture

System overview

Architecture diagram — Micro Data Centre
NETWORK LAYER   → pfSense · 10+ VLANs · DMZ, Mgmt, Tenant, Wi-Fi isolation
COMPUTE LAYER   → Proxmox VE · 8–12 KVM VMs · cluster quorum + shared storage
ORCHESTRATION   → Docker Swarm · 5-node · replicated services · health-aware scheduling
LOAD BALANCING → HAProxy · backend health checks · automatic failover routing
DATA LAYER    → PostgreSQL + Redis · replication + replica failover · NFS-backed persistence
OBSERVABILITY   → Prometheus + Grafana · live dashboards · alerting · metrics endpoints
DELIVERY     → GitLab CI/CD · validate → deploy → verify · ~60% less manual work

Proxmox Cluster

Virtualised infrastructure hosting 8–12 VMs across multiple isolated nodes. Snapshot-based recovery and shared NFS storage underpin the backup strategy.

pfSense

Routing, firewalling, and VLAN-based segmentation across 10+ networks — enforcing strict isolation between tenant, management, DMZ, and Wi-Fi segments.

Docker Swarm

Five-node container orchestration with replicated services, health-aware scheduling, and automatic rescheduling on node failure.

HAProxy

Load balancing across replicated Odoo containers with backend health checks and traffic failover — keeping the application accessible during node disruptions.

PostgreSQL & Redis

Persistent data layer with replication for durability and Redis caching with replica takeover for low-latency failover.

Prometheus & Grafana

Full observability stack: live metrics collection, alerting rules, and purpose-built dashboards reflecting cluster health in real time.

Highlights

Key features

High Availability

Replicated services, HAProxy failover, and Redis replica takeover — validated under live node interruption with zero downtime.

Network Segmentation

pfSense enforces strict boundaries across 10+ VLANs: tenant workloads, management interfaces, and DMZ services cannot cross-contaminate.

CI/CD Automation

GitLab pipeline covers validation, deployment, and post-deploy verification — cutting manual configuration effort by ~60%.

Observability

Prometheus scrapes metrics from infrastructure components; Grafana surfaces cluster health, resource pressure, and request rates in live dashboards.

Backup & Recovery

NFS-backed persistent storage combined with Proxmox VM snapshots gives point-in-time recovery for both the platform and its data layer.

Failure Simulation

Node kills and service interruptions tested under realistic load — proving the architecture holds before the grade submission.

Technical depth

Implementation focus

  1. Proxmox cluster setup: Provisioned 8–12 VMs across nodes with shared NFS storage; configured quorum and snapshot policies for recoverability.
  2. pfSense VLAN design: Mapped 10+ VLANs to functional segments (tenant, management, DMZ, Wi-Fi, CCTV); wrote firewall rules enforcing least-privilege cross-segment access.
  3. Docker Swarm orchestration: Defined Swarm services with replica counts, restart policies, and health check thresholds; assigned node roles to control workload placement.
  4. HAProxy configuration: Wrote backend definitions with active health checks and failover timeouts; validated that traffic rerouted cleanly when nodes went dark.
  5. PostgreSQL & Redis replication: Configured streaming replication for the database and Redis replica promotion — tested each failover path explicitly.
  6. GitLab CI/CD pipeline: Built three-stage pipeline: lint/validate → deploy → smoke-test; parameterised environments to support multi-tenant rollouts.
  7. Prometheus & Grafana stack: Wrote scrape configs targeting all services; built dashboards for cluster health, Odoo request metrics, and database lag.
Takeaway

What this demonstrates

  • End-to-end ownership of a production-grade infrastructure platform — from VM provisioning to CI/CD and observability
  • Network engineering depth: VLAN design, firewall policy, and traffic isolation across 10+ segments
  • Container orchestration beyond the basics: Swarm replica management, health-aware scheduling, and verified failover
  • Database operations under pressure: replication, failover, and persistence strategy for a live SaaS workload
  • Automation discipline: GitLab CI/CD cutting deploy effort by ~60% through repeatable, testable pipelines
  • Observability-first mindset: metrics collection and dashboards built alongside the platform, not bolted on after
GitHub Repository Back to Home
← Portfolio Hire me →