From Frankenstein to Focused: The 2026 Homelab

• Intel N100 CPU
• 32 GB RAM
• 500 GB internal NVMe storage
• 200 GB external SSD

• Raspberry Pi 4
• 4 Bay Icybox Direct Attached Storage
• 2 × 4 TB HDDs configured as a mirrored RAID array

• Raspberry Pi 4
• Secondary Ethernet via USB

• Netgear 16-port 1 GbE smart managed switch
• TP-Link 5-port 1 GbE PoE unmanaged switch

• Raspberry Pi 4
• USB Sonoff Zigbee coordinator

Keep

Proxmox has been reliable as a single-node hypervisor and has enabled rapid experimentation with virtual machines and containers. The performance issues I’ve encountered are largely attributable to underlying hardware limitations rather than Proxmox itself. At this point, I have no concerns with the platform and am happy to continue using it as the foundation of my homelab.

Possible Alternative: XCP-ng an open-source virtualization with enterprise features, high availability, and clustering support.

Replace

OpenWRT has proven to be a stable and capable firewall and routing platform, offering flexibility through additional services such as VPN routing. While I have no issues with its reliability or feature set, migrating to an alternative solution could help broaden my understanding of networking and firewall architectures.

Possible Alternatives: OPNsense or pfSense provides a similar feature set with a modern interface and more enterprise-grade options, while also supporting HA configurations.

Replace

Pi-hole has been a dependable service within the homelab, handling DNS sinkholing, local DNS resolution, and DHCP for the network. Its GUI is convenient and lightweight, making management easy.

However, Pi-hole abstracts many of the underlying systems. To gain deeper understanding of DNS and DHCP while maintaining manageability, a more native approach is desirable.

Possible Alternative: Webmin provides a web-based interface to manage Bind9, dnsmasq, and DHCP directly on the host. This allows full visibility into configurations and control over the native services, giving hands-on experience with the core DNS/DHCP systems while retaining a GUI for convenience.

Replace

Nginx Proxy Manager has significantly simplified TLS certificate management and internal service routing. Over time, it has become a central dependency for many services. Given its importance, introducing a highly available proxy solution would improve resilience and reduce the impact of a single point of failure.

Possible Alternatives: Traefik or HAProxy provide more direct control over routing, SSL, and load balancing while supporting dynamic configurations for modern applications.

Retire

Portainer provides a web interface to manage Docker and containerized workloads, simplifying deployment and monitoring.

While it has been reliable and useful during early adoption of containerization, it abstracts much of the underlying Docker workflow. This reduces direct interaction with core concepts such as container lifecycle management, networking, volumes, and image handling.

Possible Alternative: Managing containers directly using the Docker CLI with docker compose promotes hands-on interaction with native Docker components.

Review Later

Nextcloud provides centralised file storage with cross-platform access and external availability. It has been reliable and meets current needs. However, many features are unused; I can either deepen my use of Nextcloud’s capabilities or explore a more focused file sync solution.

Possible Alternative: Syncthing provides lightweight peer-to-peer file syncing across devices without the overhead of a full platform.

Review Later

Emby has been a reliable media streaming platform with steady development and consistent performance. It currently meets all of my requirements, and there are no immediate drivers to replace it.

Possible Alternative: Jellyfin is fully open-source, actively maintained, and offers a growing ecosystem of plugins.

Keep

Home Assistant has become deeply embedded in household operations, providing reliable automation, dashboards, and device management. Acting as the central control plane, it consumes device state and events from services such as Zigbee2MQTT while presenting a unified interface for the household.

To complement this, Node-RED can be introduced as a companion service for building more complex, event-driven automations, creating a clear learning path for event-driven architecture and system orchestration.

Deployment Options: Node-RED, MQTT, and Zigbee2MQTT can be run as separate containers to gain hands-on experience with the underlying technologies and achieve greater resilience.

Possible Alternative: OpenHAB – a vendor and technology agnostic open source automation software.

Keep

Uptime-Kuma provides service monitoring with customizable alerts and notifications. Its lightweight design and intuitive web interface make it simple to deploy and maintain. The service has been stable and effectively meets the homelab’s monitoring needs.

Future considerations include integrating Uptime-Kuma with Home Assistant automations, allowing alerts to trigger recovery actions or other automated responses.

Possible Alternatives: Prometheus + Grafana or Zabbix – more advanced monitoring and alerting, enterprise-grade insights.

Problem

The homelab currently has no UPS. As a result, it cannot tolerate short power outages or perform a graceful shutdown, risking filesystem corruption and service instability.

Resolution

Introduce a UPS capable of sustaining the homelab during short outages and providing sufficient runtime for an orderly, automated shutdown when battery capacity is low. Integration with hosts and critical services will be required to avoid abrupt power loss.

Problem

Media storage is hosted on a USB DAS that does not automatically power on after a power loss. When this occurs, OpenMediaVault starts without its backing storage, leading to failed mounts and cascading failures in media-related containers.

Resolution

Replace the existing USB DAS with storage hardware that supports automatic power-on after an outage. This removes a single point of failure and ensures storage-dependent services can recover cleanly without manual intervention.

Problem

Spikes in I/O delay and pressure stalls within Proxmox create significant performance bottlenecks, causing intermittent system responsiveness. Diagnostic tools included iotop, iostat, and systemd-cgtop.

Resolution

1. Upgrade to high-performance storage with optimized I/O throughput
2. Implement granular resource management and precise host-level controls
3. Right-size host CPU to match workload demands
4. Deploy secondary compute node for workload distribution and resilience

Problem

Although the house is wired with Cat6A, most active networking equipment is limited to 1 GbE. While the physical infrastructure is ready for higher speeds, the hardware currently caps throughput.

Resolution

Incrementally upgrade network interfaces and switching to support 2.5 GbE or 10 GbE where it provides measurable benefit. Existing cabling allows this to be done selectively rather than through a full network replacement.

Problem

When initially deployed, 4 TB disks felt generous. Over time—particularly with media and backups—capacity has filled faster than expected, resulting in reactive rather than intentional expansion.

Resolution

Redesign the storage layout around higher-capacity modern disks with deliberate room for expansion. Capacity planning should balance redundancy, power consumption, and future growth.

Problem

On rare occasions, remote management of the homelab is required. While front-end services are accessible externally, management interfaces are intentionally not exposed to the WAN. This improves security but limits the ability to respond to issues when off-site.

Resolution

Introduce secure, authenticated, and encrypted entry point for remote administration, enabling off-site management without exposing internal interfaces directly to the internet. Access should remain tightly scoped, logged, and auditable.

Problem

As my homelab has grown, new services have been deployed and the household has become increasingly dependent on it. Despite this, little consideration has been given to backups beyond the use of a mirrored RAID array, which provides redundancy but is not a substitute for a proper backup strategy.

Resolution

Introduce a backup strategy that protects critical application data and configuration. Backups should run regularly to independent storage, with retention policies that allow recovery from both recent failures and historical data corruption. The entire backup process should be automated to ensure consistency and reliability.

A key goal of the homelab redesign is to improve resilience and capability without significantly increasing ongoing power consumption.

Estimated Total Average Load: 💡 ~62.5–148.3 Watts continuous

Using an electricity rate of 23.74 pence per kWh, the estimated running cost is:

• Daily consumption: ~1.5–3.56 kWh
• Daily cost: ~£0.36 – £0.84
• Monthly cost: ~£11 – £25
• Annual cost: ~£130 – £300

Requirement: Replaceable batteries with 30–60 min runtime and USB or Ethernet monitoring

Automatic Voltage Regulation (AVR) stabilizes the mains power supply voltage to a load which provides protection from power problems.

Dataline Protection is an electrical safety measure that guards electronic devices against power surges and voltage spikes traveling through communication lines such as Ethernet.

Future resilience option:

A whole-house battery system can protect all electrical loads and dramatically extend runtime. When paired with a UPS, this ensures clean power delivery and controlled shutdown even during extended outages.

Requirement: 4-Bay, 2.5 GbE with NVMe support for caching

Requirement: Small footprint, with dual 2.5 GbE NIC for data and storage

Requirement: Dual NIC at a minimum, with enough resources for extra services

Requirement: Powered using PoE, and good range

Requirement: Energy-efficient, 2.5 GbE capable, optionally managed