So you’re building the editing NAS yourself instead of buying one. An old workstation or a fresh mini-ITX board, a stack of drives, a free weekend… and one choice that will outlive every part in the box: Unraid or TrueNAS.
Search that question as an editor and the top result is a Reddit thread from two years ago with 24 conflicting answers. The two most detailed builds ranked under it, “Video Editing NAS for 5 editors” on two different forums, are both from February 2021… back when TrueNAS meant the FreeBSD-based CORE and Unraid had no ZFS at all. Neither of those things is true anymore, and most of the advice on this matchup was built on top of them.
We’ve covered the buy-a-box path in QNAP vs Synology and the full DIY build in the editing server guide. This page settles the operating system question, and the numbers doing the heavy lifting come straight from Unraid’s and TrueNAS’s own documentation, not a comment section.
TLDR
If the NAS is your working drive, the volume your timeline reads from all day, run TrueNAS Community Edition. ZFS stripes reads across every drive in the pool and caches your hot media in RAM, which is how a box of spinning disks keeps a 10GbE line fed. And it costs nothing.
If the box is a media server first and your drives are a mismatched pile you grow one disk at a time, run Unraid. The parity array is still the only mainstream design that accepts a random 18TB into a stack of 8s and 12s with no planning, and Community Applications is still the easiest app catalog in self-hosting. That convenience runs $49 to $249, paid once.
And the correction most published comparisons miss: since January 2025, Unraid runs native ZFS and doesn’t need its parity array at all. Set the array to zero slots and you have Unraid’s interface wrapped around TrueNAS-style pools. These two stopped being opposites a year and a half ago, and almost nobody writing about them has noticed.
| Your build | Run | Why |
|---|---|---|
| The NAS is the working drive | TrueNAS Community Edition ($0) | Striped ZFS reads plus a RAM cache fill 10GbE |
| Media server first, mismatched drives | Unraid ($49 to $249, perpetual) | Parity array grows one disk at a time; easiest apps |
| Unraid’s UI, TrueNAS’s storage behavior | Unraid 7 with array-free ZFS pools | Native ZFS, RAIDZ expansion, zero array slots |
The rest of this page is the reasoning behind those three rows, plus the conditions that flip each one.
The parity array, in Unraid’s own numbers
Unraid’s classic array is not RAID. Every file lives whole on one disk, and one or two parity drives protect the set. For a media library it’s a great design: mixed drive sizes are fine, idle drives spin down, and a dead drive takes nothing with it.
For a timeline it’s a bad design, and you don’t have to take my word for that… Unraid’s own documentation puts numbers on it. Reads come off a single drive at that drive’s speed, 70 to 250 MB/s for hard drives. The default write mode runs 20 to 40 MB/s, and the summary line is even more blunt: “Without a cache drive: Average 20-30 MB/s, with peaks up to 40 MB/s.” Turbo write lifts that to 40 to 120 MB/s and needs every drive in the array spinning to do it. On growth, the same docs describe the array as “Add drives, but no striping or performance scaling.”
Translate that with our storage calculator. ProRes 422 HQ at UHD 23.98 runs 707 Mbps, call it 88 MB/s per stream. One array drive reads a single stream comfortably, maybe two on a good day. A four-angle multicam wants about 350 MB/s, and no hard drive sustains that. Writes are worse: at 20 to 40 MB/s, offloading a 500GB shoot day onto the bare array takes three and a half to seven hours.
“But people edit off Unraid for years and swear by it.” They do, and they’re not lying. They’re editing off an NVMe cache pool, which Unraid rates at 250 to 900 MB/s for writes, or they’re cutting proxies and long-GOP camera files that sip 12 to 25 MB/s, while the Mover flushes everything to the array overnight (“typically at 3:40 AM,” per the docs). Single editor, proxy or H.264 workflow, NVMe pool in front: editing off Unraid works. Multicam ProRes, DNx, or a second editor pulling at the same time: the array numbers above are the wall you hit. And past two or three editors pulling from the same box, the OS stops being the interesting variable at all… the real question becomes NAS versus SAN.
What changed in 2025, and why the old advice is wrong
Unraid 7.0 shipped in January 2025 with ZFS as a first-class filesystem, and it went further than most coverage noticed: you can set the parity array to zero slots, and on a fresh install zero is now the default. The release notes: “For new installs, the default number of slots to reserve for the unRAID array is now ‘none’.” Unraid 7.1 imports existing pools (“Import foreign ZFS pools such as TrueNAS, Proxmox, Ubuntu, QNAP”), and 7.2 added RAIDZ expansion: “You can now expand your single-vdev RAIDZ1/2/3 pools, one drive at a time.”
And here’s the detail that should retire the “Unraid bolted ZFS on” reflex for good. Current Unraid stable (7.3.2) ships OpenZFS 2.4.3. Current TrueNAS stable (25.10.5) ships OpenZFS 2.3.4. The parity-array company is shipping a newer ZFS than the ZFS company, whose 2.4 branch is waiting on TrueNAS 26. Don’t read a performance claim into a version string; there isn’t one. What it does prove is that Unraid keeps its ZFS on the same update cadence as the rest of the OS.
TrueNAS spent the same stretch consolidating. TrueNAS CORE, the FreeBSD line every pre-2022 guide means when it says “TrueNAS,” is finished: “13.3-U1.2 was the final release for the TrueNAS CORE 13.3 software train,” and its documentation hit end-of-life in April 2025. The Linux-based SCALE was renamed TrueNAS Community Edition, and the current stable is 25.10 “Goldeye” (October 2025, patched through 25.10.5 in July 2026). TrueNAS 26 is an early-adopter beta right now and nothing more. Build on 25.10 and ignore it.
So here’s where 2026 leaves you: both operating systems run current ZFS. You’re not picking a storage technology anymore. You’re picking a price, an interface, and how the box grows when you fill it.
Why editing-first still means TrueNAS
If both run ZFS, why does the working-drive verdict still say TrueNAS? Because TrueNAS is built around exactly one job, and it writes down the corners of that job that bite editors. Here are the ones worth more to you than any spec sheet.
RAM is your read cache. ZFS keeps recently read blocks in a RAM cache called the ARC, and TrueNAS states the sizing plainly: ZFS “uses all of the RAM installed in a system to make the ARC as large as possible.” For editing that means your second pass through a scene and your multicam re-reads come out of RAM at memory speed instead of off the platters. It also means the cheapest real upgrade for an editing NAS is more RAM, and TrueNAS says to max it before buying any cache SSD: “Do not add L2ARC to a system with less than 32 GiB of RAM.”
Skip the SLOG. Forum lore says every serious ZFS build needs a fast SLOG device. TrueNAS’s own primer: “Synchronous writes are relatively rare with SMB, AFP, and iSCSI, and adding a SLOG to improve the performance of these protocols only makes sense in special cases.” Your editing clients talk SMB. Put that money into RAM or another drive.
And don’t fill the pool. From the same primer: “At 90% capacity, ZFS switches from performance- to space-based optimization, which has massive performance implications. For maximum write performance and to prevent problems with drive replacement, add more capacity before a pool reaches 80%.” Size for that from day one… a “48TB” pool is a 38TB pool if you want the throughput you built it for. The calculator turns the difference into shoot days per codec.
One more vendor rule for anyone buying today’s big drives: “RAIDZ1 is not recommended for drives over 1 TiB in size.” That’s TrueNAS’s line, and a modern 20TB NAS drive is twenty times the threshold, so plan on RAIDZ2 and budget the extra drive. Picks are in the NAS hard drives guide.
Growing the box: still Unraid’s home game
Expansion is where the parity array keeps a real advantage, RAIDZ expansion or not. The array takes any drive up to your parity size, one at a time, mixed ages and capacities, no layout rules, and every added terabyte is usable. Seven drives from three eras is a normal Unraid array. It’s the whole reason the design exists.
RAIDZ expansion, which both platforms now ship, is narrower than the headline suggests. You add one drive at a time and the pool stays online, but TrueNAS’s pool docs list the fine print. The parity level never changes: “a four-disk-wide RAIDZ2 expanded to a six-disk-wide RAIDZ2 still cannot lose more than two disks at a time.” Existing blocks keep their old data-to-parity ratio, so “an extended RAIDZ VDEV can report a lower total capacity than a freshly created VDEV with the same number of disks.” And the incoming drive has to be at least as large as the smallest one already in the vdev, so mismatched sizes still waste capacity. You can’t shrink, and you can’t remove a drive.
Unraid 7.2 also added an editor-shaped trick that nothing ranking on this search mentions: the array now accepts drives already formatted NTFS, exFAT, or ext4, data intact. Unraid’s docs pitch the exact scenario: “say you are a content creator with a box full of hard drives containing all of your historical videos”… add them to a new array and “the data drives will retain their data.” No copying, no staging drive. The one-way door: do it before you add parity, because “once a parity drive has been added, any new data drives will be zeroed out when they are added to the array.” If your archive is a shoebox of full drives, TrueNAS has nothing like this.
What they cost
TrueNAS Community Edition is free. Not freemium: the full OS, every feature, no device caps, on any x86 hardware. The download page puts a newsletter signup in your way, and you can skip it.
Unraid is a paid perpetual license, and “perpetual” is the part that gets mis-reported everywhere. The buy page’s own headline is “One Platform. Every Feature. No Subscription.” and the licensing FAQ answers “Is Unraid OS a subscription?” with a flat no. Three tiers: Starter at $49 (six storage devices total), Unleashed at $109 (unlimited), Lifetime at $249 (unlimited, updates forever). Starter and Unleashed include a year of OS updates; after that you can pay an optional $36 a year to keep getting them, or don’t, and keep running what you have.
Two footnotes before the sticker price settles it. The Starter cap counts every storage device, so six hard drives plus a two-stick NVMe cache pool is already over the line. Most real editing builds are $109 builds. And skipping the $36 has a cost: Unraid supports only the two most recent minor releases, so a lapsed key falls out of security patches within a couple of versions. Your install keeps running fine, but the security fixes stop arriving.
Notice what the money buys, though. TrueNAS gives away the storage engine, and Unraid now ships the same one. The $49 to $249 covers the interface, the app catalog, and the parity array. Whether that’s worth paying depends on how much you want those three, because none of that money makes your storage faster.
Neither OS will fix your network
Most DIY editing servers hit the wire long before the OS becomes the limit. Unraid’s own docs note that a 10GbE network caps around 1,100 MB/s, and no OS choice moves a gigabit link past about 110. If you’re still on gigabit, that upgrade beats every tuning trick on this page; the 10GbE guide covers switches, adapters, and when 2.5GbE is enough.
Once you’re on 10GbE, SMB is where the two projects part ways. TrueNAS ships a dedicated SMB multichannel tutorial with the failure modes named: “SMB multichannel does not function if you combine NICs into a LAGG” (the most common self-inflicted wound in team setups, since two bonded 10GbE ports feel like they should double your speed) and “Do not configure multiple SMB multichannel interfaces on the same subnet.” Unraid publishes nothing equivalent. A search of docs.unraid.net for multichannel or 10GbE tuning comes back empty; the SMB docs cover shares, security, and Time Machine, and throughput tuning happens on the forums. Unraid 7.3.2 ships a current Samba, so the capability is there. What’s missing is the vendor writing down how to make it fast, and that gap costs you an evening of forum archaeology the first time multichannel doesn’t come up.
Apps, Plex, and the media-server case
Unraid’s Community Applications is still the friendliest software catalog in self-hosting: one-click templates for Plex, Jellyfin, the *arr stack, backup tools, and VMs with GPU passthrough in the same UI. If half the reason for this box is serving the family movie library, that weighs as much as any throughput number.
The old counterargument, that TrueNAS apps were a Kubernetes science project, expired in late 2024 when TrueNAS moved its app system to plain Docker Compose. Catalog apps install through a wizard, and any public Docker image deploys from a YAML file. Unraid’s catalog is still easier. TrueNAS’s is no longer a reason to walk away.
One upgrade warning for the Plex crowd: TrueNAS 25.10 switched to NVIDIA’s open kernel drivers, and “Pascal, Maxwell, and Volta architectures are no longer supported.” The old GTX 1050 or 1060 doing your transcodes stops doing them on upgrade; GTX 16-series, RTX, and newer survive. For the wider question of sizing one box for both transcoding and editing, the editing server guide covers the Plex side of a shared build.
The same release also retires two things upgraders lean on: built-in SMART test scheduling, which gets migrated to cron tasks on upgrade, and the SMB recycle bin on new shares. TrueNAS points you at ZFS snapshots instead, which is the better tool for a deleted project folder anyway. Our archiving guide covers snapshot-and-replicate as a backup plan rather than a panic button.
In conclusion
Run TrueNAS Community Edition when the NAS is the working drive: striped vdevs and the ARC are what fill a 10GbE line off spinning disks, the documentation tells you the failure modes before you hit them, and it’s free. Run Unraid when the box is a media server that also holds projects: mixed-drive growth and the app catalog are worth $109 to the right builder, and the NTFS-adoption trick alone can justify it if your archive lives in a shoebox of full drives. And if what you want is Unraid’s interface with TrueNAS’s storage behavior, that stopped being a compromise in 2025: array-free ZFS is Unraid 7’s default for new installs.
If this whole page reads like a second job, that’s an answer too. Buy a QNAP or Synology and let the firmware make these calls. DIY storage pays off in flexibility and dollars per terabyte, but only if you enjoy the tinkering.
And that’s it. This is how I’d split the choice with my own drives on the line; your codecs, team size, and appetite for maintenance may tilt it. Every number above comes from the two vendors’ own docs as of late July 2026, and both projects ship fast, so skim the release notes before you commit a weekend. Questions about your build? Leave a comment.
