▲ 198 points
back
171 comments
This architectural shift eliminates the user’s ability to expand memory capacity as self-hosting demands grow, effectively putting a permanent ceiling on running memory-heavy Docker containers, ZFS pools, or virtual machines.
That's not a NAS anymore, that's a full server. I think of a NAS as "low-power computer with lots of storage attached", so not being able to upgrade the RAM on one isn't all that surprising.
For business it should be all storage. But for home use the best is all-in-one. A home user doesn’t benefit from specialization of equipment, just brings them more headaches managing more stuff. One device that has storage and can also run Plex, Immich, HA in a single package makes the lives much easier for those who have to take care of the IT in the house especially of they’re not “IT people”.
Well that's just for home use. Say a small to medium enterprise needing a shared drive. Or even things like say video editing you might need a 10G connection. Furthermore for most it's just a hobby so there's no right or wrong, if you can live with the noise and power of old enterprise gear then kudos to you, or if you think a raspberry pi with a sata hat works then go ahead
> That's not a NAS anymore, that's a full server. I think of a NAS as "low-power computer with lots of storage attached", so not being able to upgrade the RAM on one isn't all that surprising.
Home NAS machines have often been used as full servers and in the past allowed those upgrades. And it makes sense in a home to have one device that's flexible enough to fill multiple roles, be small, and quiet.
I think of a modern consumer NAS as a backdoor server that a user buys to solve one problem they know they have, plus a whole bunch of additional problems they didn't know they have... A critical part of of the vanguard in the war against proprietary centralizing webapps.
If I have CPU power at my disposal, why should I waste it? I love the fact that my NAS can serve my other needs too.
I think we can all agree that "NAS" for consumers is a marketing term, like "atomic clock". Anyone who works in an office with a NAS may come to admire its functions, and want the same experience at home.
https://en.wikipedia.org/wiki/NetApp
When I think of NAS, I think of Network Appliance. Even in the mid-90s they were already well-known as the makers of the most formidable turnkey NAS that an enterprise could have. They are so awesome and specialized that I never even worked near one. Their systems were called "filers" and they have proprietary filesystems with proprietary kernels and CLI, that present with an NFS interface, among other things. The Filers are blazingly fast and ultra-high-capacity.
I run some light services on my NAS cause well, in the end, the distinction in naming is arbitrary at best, it’s a computer, and I use it for computer things. In an ideal world where I can afford separate purpose-built machines for everything, I’ll have a storage-only NAS, but it’s a computer I have that also happens to have many hard drive bays. Being able to service my own machines shouldn’t be all that surprising either.
Then, sorry, but you don't understand today's NAS market. All it takes is looking at the products QNAP, Minisforum, TruNAS, UnRAID and many more have been putting out for years - and it sells well. Nobody wants more infrastructure than they need and a well powered NAS is a great place for homelabbers and SMB to invest.
I just recommended an N5 Air [0] to a friend because it's got a powerful CPU, can expand to 96GB of RAM, has a great mix of NVME and HDD and has a PCI slot they can run an Intel A310 in for local transcode. It's the perfect Proxmox host to run everything they'd want to do including local services but also fault tolerant storage for a reasonable price.
You missed the point the author really complains about: soldered in memory chips create a single point of failure.
What you quoted needs to be seen in the context of what author calls "increasing media demands". I.e. the author believes that while the way the system is configured might be OK today, it won't be OK in the future.
> That's not a NAS anymore, that's a full server. I think of a NAS as "low-power computer with lots of storage attached"
I disagree. And we can at least discuss whether a NAS that doesn't have ECC is that much of a NAS or not.
I'd take an old Xeon server with 64 GB of ECC memory, run its CPU in powersave mode, and slap TrueNAS (with ZFS) on it any day over buying a dedicated NAS.
YMMV but actual professional server gear, with ECC memory, is to me more of a NAS than a consumer NAS.
And nothing forces you to run a "full server": you can dedicate a real server to be only a NAS and slap the lowest consuming Xeon on it, which moreover you can run with a powersaving CPU governor.
Slap a 10 Gbit/s ethernet NIC in it and call it a day.
It shall be more reliable too.
You can of course cosplay with a consumer NAS but the real thing is better.
P.S: mine is a HP Z440 with a 14-cores Xeon (so a workstation, not a server: the advantage is that it's totally quiet)
I built my own NAS, just Ubuntu+ZFS using an old PC I had and it works great. Don't know why people are so crazy about buying prebuilt NASes.
Why wouldn't they be? No fiddling, super compact form factor, (usually) far superior energy efficiency, maybe phone apps, did I mention no fiddling?
Also, the form factor!
Its the maintenance and details that kill your NAS.
You will probably rely on it for 10 years at least, so building your NAS means committing to maintaining it in the long term, you have to be sure this is something you will uphold.
Also while setting up your NAS initially is definitely doable, there are a trillion little things to get right and make sure won't bite you later on (scrubble, etc).
All these things compounds if the NAS is not just for you, but your wife, kids, family, etc. Now it becomes your responsibility that all these precious data and photos entrusted to you are 100% safe for the decades to come.
For me, it was laziness. While I have the skills to built one from parts, I just wanted a big drive in an aesthetically pleasing box. I didn't want a project. I wanted something small and somewhat rugged. So I got a synology 5-bay almost seven years ago and have never looked back. It lives beside my UPS and wifi router in a small tower, all objects I expect to "just work" for months on end without any attention.
My AI rig ... that is a project. (After a kernel update, one of my GPUs has disappeared. Fixing that will be a project for my next day off.) I don't want my NAS to ever become a weekend project.
Many of us are sysadmins at home. Home Assistant, home NAS, home DVRs, etc. Consumer NASes are in tension against the DIY hacker / homelab people who often run these things. These NASes are just complicated enough to be consumer unfriendly, but not enough for "our" needs. The author has a good point about how NAS hardware is more integrated - less capable for power users - while the software has stagnated into a sort of semi embedded, minimal viable product.
Just yesterday I heard about a friend-of-a-friend type situation where the sysadmin husband was in hospice, and the wife was struggling to figure out how to operate the NAS to recover family photos, etc.
I think consumer NAS manufacturers could add a "foolproof backup connection method," such as a USB port in USB device mode, so the NAS could be easily connected to a laptop by someone with little computer knowledge, in read-only mode.
That said, my home NAS is "the littlest server": an aging ProLiant DL380 Gen8 MicroServer. It's behind a double conversion UPS. SAS card + drives, Proxmox running virtualized FreeNAS, etc. Typical overdone geek build. I have not seen a finer NAS platform for the precocious home admin. I wish there were newer versions which kept the enterprise style "lights out management," and added m.2 and such, but I think HP Enterprise as a company is not interested in the homelab use case.
The cookie preference pop up on this site is unacceptable - 1 click to accept all cookies, SEVEN clicks to reject all cookies.
I don't share the author's view of the OSs becoming copycats of eachother and not that specialized.
Truth to be told, a NAS is the perfect place to have your home server. You can have both your applications and data on the same device and avoid having to build a lab just to have nextcloud or some other utility self hosted. However, to achieve this, the OS must become more similar to a normal linux distro, for better or for worse
While I don't disagree with the individual complaints, I've seen this niche go from running scuffed OSes on underpowered Synologies or building DIY NASes with limited case options like the Node 304, to multiple hot swap NASes running great x86 chips (and memory slots) and many DIY build paths including 3D printed cases. Things probably suck for those reliant on Synology, but I'm stoked for this hobby, especially when Minisforum/Aoostar to start pumping out Wildcat Lake stuff.
Given the (non-) responsiveness of the page, I’m going with they focused too much on “storage” and not enough on “network attached.”
Just last month I bought a HP Micro server (N54L) from 2012, two second hand 4TB drives, 16GB of DDR3, and a 2.5GbE dual port network card. All in, that was about £180. I may shell out another £30 or so for a half depth, low profile graphics card that can support a couple of transcoding streams.
Reading the article, I don't regret it. Part of what puts me off the regular NAS market is price, but also things like non redundant boot options, ambiguity about whether or not I can install my own OS, and the lack of slots. The NAS form factor itself is off-putting because if you're buying something that isn't Synology or whatever, the RAM tax is still there.
Maybe I'm a cheapskate who will have slightly higher opex, but old hardware does seem to be pretty much as good as new hardware, to me.
Expensive drives have become such a pain. Maybe five years from now, if supply meets demand, I just wish I can afford a 1PB all-flash NAS. I really don't want to hear noise and I wish data centres would release their used supply on market. I will put backup on HDDs.
Soldered RAM and novel format hardware is a pain point. Macbook storage upgrade needs special custom hardware. Synology still doesn't allow third party HDDs in their extension units. NAS devices are truly a fantastic category on its own. My data remains local, while also allowing me to use it as a server for my workloads.
I use a mini amd pc with a disk bay. Surprisingly it is cheaper than many nas, and much more configurable (from memory to ssd) and have a much better gpu.
The only challenge would be finding a very good cable that stays solidly.
I agree on most points, except perhaps for the expectation of compute and RAM since NAS = ‘network attached storage.’
I’ve had a HP N40L for 15 years and every time I see the current offerings I hope I don’t have to look for hardware soon; it’s been such a great homelab server. After FreeNAS > TrueNAS Core, I now roll my own FreeBSD. It felt like a great leap into the deep end going from GUI to CLI, but I haven’t once looked back.
I didn't see them mention anything about adding features that aren't strictly about storing files... things like containers and virtual machines...
On the first point specifically, everyone is going to need to wrap their heads around memory being more tightly integrated with compute in future generations. The modular form factor restricts DRAM speed, bus width, hurts reliability and cost. Memory modules provide flexibility, but the cost is real. I suspect in the future we will get better at re-balling memory ICs.
It's easy to build one yourself. Just don't buy from vendors if you can't run your own software. It's that simple.
I've run two NAS appliances for 15+ years now, and never had to worry about any of this, because I just run good old Community Linux or BSD on it.
I can plug whatever PCIe card I want, as much storage as I want and from whom I want.
I think the biggest issue is hard drive (or SSD) costs currently. I got a UNAS 8 Pro a few months ago and it still sits empty. It's my own fault, I should have priced out the drives but I was buying Unifi equipment for a new house and wanted to try our their NAS offering.
I just looked and the hard drives I've bought before are currently >3x the price I last paid. I know I'm not telling anyone anything new but I'm just praying my existing drives hold out since replacing them (even 1) if they fail will cost a small fortune.
Here's my single story with NAS in 10+ years of software development.
A client had a NAS for their SaaS startup. Disks were having issues. The NAS was a prosumer product that was connected via USB. Had some SATA to SATA cable shipped by Amazon to the garage/colocation facility which basically meant removing the NAS and connecting the disks straight to the motherboard.
Issue solved.
It seems like one of those tarpit product categories that you are just better off by skipping completely, like ORMs, MCPs or reverse proxy HTTP servers.
I built my "NAS", which is just a normal computer with no GPU but 3 hard-drives in RAID 5, for $400 + storage cost. It is much more powerful that any NAS, even thousands of dollars ones, I'm not too sure at what cost (power consumption maybe?).
I don't understand why people would go with an actual pre-built NAS from a manufacturer over my approach. It's always a complete rip-off.
Are there any good out of the box NAS options?
You can build your own a lot more cheaply now especially if you only want storage. There’s some canned distros and control panels for it.
If the storage doesn’t need to be fast a newer Pi can do it with USB-C drives.
An old Intel Mac mini from eBay makes a good home file server too and is compatible with everything. Get a thunderbolt drive bay.
I don't think it is that bad. Especially Ubnt just released their UNAS Pro. ( Although only comes with BTRFS on consumer and ZFS only on the Enterprise. )
I still think Apple should have bring their AirPort Express and Time Capsule back. Instead they wanted everyone to be on their iCloud subscription.
I am rather surprised that there's no well-supported OSS way to get full disk encryption + ZFS + automatic SSD/HDD tiering or at least caching.
FDE with Secure Boot on Linux is a minefield.
ZFS doesn't do tiering.
Sadly on Windows tiering is also expensive (requires Server editions).
> Software Homogenization and Interface Uniformity
how is this a negative? I would say, please do more like ugreen and support alternative OS, provide drivers and integrations.
On 1.4 "Integration of Fixed Low-Capacity OS Drives"
eMMC bad because un-upgradeable and/or repairable. m.2 bad because if it wasn't there a user could put a big storage device there.
I'm not even sure what the author wants here?
Synology is truly dead to me now. I used their stuck mostly happily since my first DS412+ way back when. But this thing with locking down features to require their own branded drives is madness. No way, no how.
Last year they tried restricting the OS from creating new volumes unless all drives in it were Synology branded. People tore them a new asshole in the press and they backed away from that.
Now their newest DSM 7.4 release features a new smart storage system with postprocessing deduplication and compression. I tried to enable that on my own volume for testing, but nope: it only works on “pure Synology” volumes.
Fuck that. Synology, get this out of your damn heads. I will never, never buy your stupid, overpriced drives. It won’t happen. Never. And on top of the ludicrous price premium, if one breaks, how do I can get a replacement urgently? If my IronWolf blows up, I can drive over to Central Computer and come home with another just like it. Amazon has few Synology drives in stock and none of them are large. When I have a dead drive, the last thing I’m gonna want to do is go on an urgent quest for a replacement unicorn.
I’m done with ‘em. It’s clear they’re fixated on this stupid idea and aren’t going to give it up. Well, good luck with that and your remaining customer base, which no longer includes me.
> Software Homogenization and Interface Uniformity
While I do prefer the DIY route, I'd struggle to call that "enshitified". Vensors consolidated around solutions because they work reliably, without reinventing the wheel. I'd much rather a NAS be a lightweight wrapper around a proven and easily maintained base than try to over extend themselves and jerry-rig some abomination.
And for customers who rely on them it means they can shift brands and still feel basically just at home.
The primary problem is the cost of drives now IMO.
I've got 16tb drives (in a usb-c DAS, not NAS), and at 2-3 the price that I bought them for I don't plan to replace them when they go.
The data will just be lost, like tears in rain.
FYI the article is unreachable. Clicking the link results in an erroneous "you are on a black list! Email admin@nascompares.com if you think this is a mistake." message.
The term got enshittified. Compare the definition over time via Wikipedia. Technically, NAS by the original definition can not host anything but a filesystem. It is simply free disk space, a SAN with a GUI. Anything you do to that disk space does not change what it is, until you start adding OS features, monitoring features, Wi-Fi, etc. Now it no longer is NAS - it is a server. We did this to ourselves.
https://en.wikipedia.org/wiki/Network-attached_storage#Descr...
https://en.wikipedia.org/w/index.php?title=Network-attached_...
I can easily accept that arguments like "an Iphone is just a computer" are often overly reductionist.
But people justifying much of this segmentation?
Come ON, any NAS is just a server. What are people DOING here?
I've always had a relatively bad experience with NAS hardware. Not necessarily due to anything the hardware manufacturer could have done. I have lost entire enclosures to lightning strikes on more than one occasion. Everyone thinks their IT skills are hot shit until they wake up to a closet full of burned hardware in the morning. It is quite a humbling experience if you've never been through it before. There isn't any kind of semiconductor process tech, file system design or RAID configuration that can save you from this. An off-site backup is the only thing that would count, which brings me to my next point.
These days I just store my files in S3. Whether or not this actually costs me more money becomes a very irrelevant shade of grey once I factor in the value of my time and sanity. I prefer having the capability to store and retrieve large files, not spending time maintaining this capability. This is also my "off-site" backup. I am paying for my operational NAS and its backup effectively at the same time. No one doing DIY is remotely close to 99.999999999% durability, even with all the kings horses and men put together.
I often see this urgency with total control pushing tech people into the very systems of oppression they seek to avoid. Putting your bits in some blob storage provider like AWS is not "losing the game" unless you frame it that way in your head. You can frame it as "getting one over on the bad guys" if you are a little bit clever with how you approach the problem. You can compel Amazon host any kind of bits with a sprinkle of AES256 on the client. I've been a customer with them for over 15 years and not once have they rug pulled me. I've never seen a NAS-style solution survive that long without intervention.
Speaking of enshitification, this blog is an unreadable mess of ads and popups. As fast as I was in I was out.
nitpicker here: enshitified or enshittified?
“NAS” itself is an enshitification
Published on a site that is enshittified with ads and nags, how ironic
Soldered RAM being bad is an opinion, not a fact.
I’ve seen way too many RAM slots go bad.
Smart people always get the highest configuration in computers, unless they know they’ll never need some of the resources and can tolerate a lower option.