Scott Tompkins

TrueNAS Storage

3 min read

TrueNAS Storage

TL;DR: TrueNAS is the storage backbone of my homelab—think of it as a smart hard drive system that automatically detects and fixes data corruption, takes instant backups, and serves files to everything else on my network. I run two nodes: one with fast SSDs for things that need speed, and one with large spinning disks for bulk storage and backups.

TrueNAS anchors the storage layer of my homelab with two purpose-built nodes, each optimized for a different workload profile. This setup has taught me more about storage architecture than years of using managed cloud storage services.

The Nodes

TrueNAS-SSD (Goldeye)

Fast storage tier backed by solid-state drives. This serves latency-sensitive workloads: VM disks that need fast I/O, databases, anything where storage latency is visible in the user experience.

The interesting capability here is NVMe over Fabrics (NVMe-oF) using NVMe over TCP. I use the TrueNAS Proxmox VE Storage Plugin to integrate automatically provisioned NVMe-oF targets directly into the Proxmox storage layer.

Why this matters: High-performance block storage used to require expensive SAN appliances with proprietary software. Now I get automatically provisioned NVMe-oF targets using open-source software on commodity hardware. The performance is excellent, and Proxmox consumes it natively—no manual iSCSI target management.

TrueNAS-HDD

High-capacity tier on spinning disks for bulk data: backups, media, archives, anything where capacity matters more than latency. This is the “cold” tier that complements the fast SSD storage.

Why TrueNAS and ZFS

I’ll admit I previously dismissed ZFS as experimental or esoteric. That was wrong. ZFS is one of the most battle-tested storage foundations in the industry. It underpins enterprise storage appliances, cloud providers, and mission-critical systems that most people don’t realize are built on it.

What ZFS gives me:

  • Checksumming: Every block is checksummed, and corruption is detected automatically. This matters—silent data corruption is real, and most filesystems don’t catch it.
  • Snapshots: Point-in-time copies that are nearly instant and space-efficient. I snapshot frequently and can roll back easily.
  • Compression: Transparent compression that’s usually a net win—faster I/O because there’s less data to transfer, at negligible CPU cost.
  • Replication: Built-in replication that I can use for backup or disaster recovery.

What TrueNAS adds:

  • Sane defaults that make ZFS approachable
  • A web interface for management tasks
  • Snapshot scheduling and lifecycle management
  • Plugin ecosystem for common services
  • The NVMe-oF integration I mentioned above

What This Has Taught Me

Storage is more interesting than I thought. When storage is invisible (as it is in cloud environments), you don’t think about it. Running my own storage taught me about durability, performance characteristics, failure modes, and capacity planning in ways that reading documentation never could.

Open-source infrastructure is production-grade. TrueNAS and ZFS aren’t second-rate alternatives to commercial products—they’re the foundation that many commercial products are built on.

Tiered storage makes sense. Not everything needs fast storage, and not everything can fit on fast storage. Having explicit tiers forces me to think about data placement, which has made me a better architect.

The Connection to Professional Work

The professional Proxmox + Ceph infrastructure uses distributed storage (Ceph) rather than the standalone ZFS nodes I run at home. But the concepts transfer: tiered storage, backup strategies, failure mode planning, capacity management.

The homelab gave me intuition for storage that made the professional work easier to design and operate.