Toshiba MG08ADA800E , Toshiba N300 8TB or Seagate Exos 7E10 8TB probably
$169.99 at current list prices
$171 at current list prices
WD Blue 8TB is CMD, so it stays as cheapest viable driv then 
You can always âshuckâ WD Easystores.
Itâs also unsuitable for NAS usage soâŚ
Does depend on your workload, if it is purely for archive/bulk storage, chances are even the SLOOOW spinning drives will be more than fast enough to saturate your network, especially using multiple spindles with ZFS.
Consumer >=10 gig networking unfortunately isnât quite a thing for most people and even a single spindle will outrun 1 gigabit ethernet for bulk streaming data storage. 4 hard drives will easily outrun 10 gigabit networking for sequential streaming IO (and ZFS will serialise writes and use read-ahead and caching for reads).
If youâre doing VM storage, etc. then sure - SSD is the way to go, but if its just a data dump location⌠rust will give you a lot more capacity that is âfast enoughâ (i.e., will mostly be network bottlenecked, especially in a single/small number of users home setup).
2 Likes
Use a zfs intent log device. The smallest and fastest disk you can find. It basically mitigates lots of fragmentation problems that arise from sync writes from stuff like NFS or VM disks. Mind you that this might be outdated information but having a log device is never a bad thing.
On the other hand ARC level 2 is not always desirable.
I hear you, my point is, if you are looking at a 4TB SSD and a 4TB HDD and the HDD costs more than 50% of the SSD, there is pretty much zero point in buying the HDD. At least for drives below $200. Even with the 8TB HDDs vs 4TB SSDs, the fact that you can fit four 4TB m.2 sticks in the same physical space as one 8 TB HDD makes the HDD a tougher sell.
HDDs are more and more losing the price per GB battle with SSDs and $200 16 TB SSDs are just around the corner. Might be two years away or five, but SSD storage will finally overtake HDD storage and it will happen a few years before 2030.
While the writing is on the wall, I can agree that we are not there yet and it will take at least 5 more years before the complete death of HDDs. I think investing in a QNAP or similar case today over something like the Flashstor is a mistake, but if you already invested in HDD infrastructure then of course use that until it stops making sense.
HDD storage is not unusable, just big, noisy and slow. So use what you have but plan for the future, same as getting an electric car - buy one as your next one, donât throw away what works great for you today
I actually agree mostly, but 4TB isnât a huge amount and chances are you have more SATA ports than you know what to do with and less m.2 slots, especially in cheap or old hardware the typical home user will repurpose as a ZFS nas.
16-20TB rust isnât that expensive and means you can not even think about how many backups to keep etc.
SSD is definitely cheap now but I think rust will be around quite some time yet as consumer networking just isnât fast enough for the SSD performance to matter for archival.
Maybe when 20-40 gig is commonplace it will be different but most home users arenât even bothering with wired networking so I doubt the price is going to come down on consumer grade (read: quiet) high speed Ethernet any time soon.
Again though; if you plan on running VMs on your nas donât even think about it. SSD all the way for those.
2 Likes
True, then again, your dad only uses his computer for word processing and email, so why did you buy him a 6 core 32 GB laptop? After all it cost a whole $100 more to that Duo core celeron machine with 8 GB of RAM, that would have been perfectly adequate.
At some point it is no longer worth it, and on the low storage needs end, I do think we are close, or even at that point now. As seen above, the same money buys you a six bay 24 TB m.2 NAS vs a five bay 40 TB mech drive. Except for 16TB less raw storage the m.2 NAS is better in every conceivable way.
Yeah, but the 16 TB of space is noticeable, for NAS archive via 1 or 10 gigabit ethernet for a small number of users, the speed probably isnât.
I agree, at some point thereâs going to be âmore space than I could ever needâ vs. âway faster but cheap enough for what I do needâ with rust vs. SSD, but Iâm not quite sure weâre there yet.
If affordable fast networking becomes commonplace (i.e., 40-100 gig) in the home - maybe. But then weâre at the point where thereâs no consumer end-user application that really needs that level of performance. Yet.
I went from HDD to SSD for media network storage primarily because the HDD would spin down (good) and take 20 seconds to spin back up when I wanted to access it (bad).
Do proper NAS boxes mitigate this somehow? Otherwise SSD NAS isnât just about raw speed but how quickly it can go from low power state to delivering data.
True, but more than likely 8 TB will hit $200 within the next 18 months, and 16 TB will hit $200 within the next three years.
Why would you even consider an HDD setup in 2026, unless for replacement parts, if they are both at price parity?
Agreed. We are not there yet but it is a lot closer than I think most people realise - and that is a good thing.
Today, there are very few reasons to invest in HDD infrastructure over SSD unless you need the capacity. You will pay a small premium for the SSD now, true, but at the same time that SSD server will be able to take new mainstream drives in 2035. By then HDDs will be hideously expensive and hard to come by just like IDE drives are today.
Anyway, I think we both mostly agree that it will be a matter of what is cheaper in the end, I think SSDs are now cheap enough that paying the extra 20% is worth it for low capacity options, and you are of course free to disagree. There are good arguments for both sides and time will prove which is ultimately the better move
I just recently bought (6) used HGST 6TB drives and put them into RAIDZ2 so that I can lose two and still be OK. Iâm hoping that wonât be the case but if it happens I have a cold spare available as well. 24TB, some peace of mind, and it wasnât too expensive since they were used. As for expansion, Iâm a huge fan of adding an additional ZFS pool or RAID array with MergerFS and then using that virtual disk from there. For me, I am thinking about taking the (3) 4TB disks I had been using and putting them into a ZRAID1 pool then MergerFSâing them together for a total of 32TB with redundancy in the form of two separate ZFS pools of 9 disks total. BUT, thatâs just what I decided to go with since I really like MergerFS and it allows me to ZFS-pool what I have in ways that make sense while still ending up with one single share in the end.