My NAS is reaching the end of its design. Twenty terabytes usable across five bays, a few hundred gigabytes left, and it’s warned me twice now that I’m nearly out. I’ve bought time where I could, moving what I can to the pool meant for cold archive. But the design is aging out on more than one front. Swapping the drives to bigger ones won’t fix the root of the problem. With the new home lab feeding the queue, the need only grows until the backlog clears. The clock is ticking.

Fortunately, with the blessing of the algorithm, YouTube started showing what’s possible outside of Synology, outside the pre-built NAS realm: more power, more storage, and somehow cheaper — the holy trinity of a better NAS, they say. All you have to do is build a PC, install a NAS-level OS, and hook the drives up. At a fraction of the price, you’ve built your first DIY NAS. They offered something from which I couldn’t look away. So I dug deeper.

Finding the right parts was no easy feat. I needed a chassis that wasn’t a regular PC tower or a server rack, but a proper case with HDD bays. To drive those bays, the motherboard had to come from a limited selection. Then a compact, high-efficiency PSU for 24/7 operation, plus CPU and RAM. All things combined, the total came to roughly half the price of the 8-bay DS1825+, for a 9-bay build. Worth mentioning: this is before shipping and customs.

Practically speaking, even if I’d placed the order at that half price, I doubt the parts would have arrived by the end of the week. The components for a power-efficient, 24/7 build with no bells and whistles are a niche item, rarely in stock — and the ones I could get can’t drive nine bays. The showcase builds often lean on power, the computing power, and not bay expandability or data integrity like ECC. So the “power” in the holy trinity really just means running some other server task — and that task, plain computing, is something a mini PC from Amazon will handle on any flavor of Linux, for a fraction of the build’s cost.

The number that should actually drive a NAS decision isn’t the unit. I’ll go further: treat it as expendable. Computing’s beside the point — it’s the drives that eat the budget. Take the DS1825+ again. Eight bays. If you’re not migrating from an existing Synology pool, that’s a set of drives sourced from somewhere else to make full use of it. And if you already have storage in some other form, you still need a starter set just to begin moving data across. The argument goes that DIY comes out cheaper — but the drives cost the same whatever box they go in, and the entry price into a new ecosystem is brutal. The drive cost is platform agnostic.

Back in the planning phase for my five-bay NAS, I meant to run 4TB drives across the board. The rebuild time for a 4TB was manageable, and SHR-1 across five drives was within my risk tolerance. Being expandable was part of the design — I didn’t start with all bays populated. The expansion happened slowly. But three things have changed that made me realize it’s nearing obsolescence: a. 8TB has outpaced 4TB on price-to-storage; b. an 8TB rebuild is past my risk tolerance; c. the 4TB drives I’ve pulled still work despite their age. So replacing a dead 4TB means paying more for less, and going 8TB means every rebuild leaves the pool exposed longer than I’d like.

Had the market not been in flux from the AI bubble, had the parts been available, I’d have considered giving Unraid a go with the build. Its real strength is the one prebuilt NASes don’t offer and the guides won’t credit. Unraid offers up to dual parity, like RAID6 or SHR-2, but it gets there without striping: each drive stands alone. So it’s not real RAID, and that’s the point. It degrades more gracefully, and the user keeps more control over the pool. A drive that’s been decommissioned but still works can get a second chance as long-term cold storage, which is exactly where my aging 4TBs would have gone.

For me, the case was settled the moment I saw the final price tag. It was either start a DIY project with no end in sight — if only it had been a week long — or pay roughly the same for a prebuilt and be done. I chose the devil I know over an adventure with 20TB of data riding on it: better throughput, better redundancy than my current setup, and real stock availability. DIY NAS is a genuinely interesting lead, especially for home and small offices. But the bigger it gets, the less room there is for creativity. Something is boring for a reason, like how old-school networks separate computing from storage.