Doing some calculations right now, and from my count I have about 16TB of data on various drives. By “various” I mean 4 separate drives, all attached via USB to a 2012 Mini that’s acting as a server.
That’s a lot of drives, and a lot of cables. It looks like the most cost-effective drives right now are about 4-8TB depending on the sales, so going with one huge drive is (a) not great from a CPOF perspective (important data could be stored on multiple drives, for example, if I use multiples) and (b) still presents the problem of needing more cables and such for expansion. Given that drives don’t run well on hubs, this is probably the biggest deal.
Wondering how best to handle this without dropping a ton of money on a Synology or something like that. There’s no need for mirroring / striping, and I actually prefer to not do RAID at this point as I have robust cloud backup, and I don’t want to functionally double the hardware cost of my setup.
Synology’s are not that expensive and especially they run forever. Mine turns 9 years old in spring, and it runs 24/7. Most robust piece of gear I ever owned, and it still gets the latest updates. They’re extremely easy to expand and you can repurpose them to do dozens of things.
However, in order to facilitate cloud backups and Plex transcoding, my next server will be a Mac Mini with an OWC DAS.
So… make your choice between two schools of thought
With a Mac Mini acting as a server, the OWC Pro Quad looks like a nice solution for your use case.
I am in the Synology camp. The Synology is my server. I am a strong believer in RAIDs, the Synology is partially being backed up to the cloud (daily), Synology Drive does a fantastic job with versioning (at least as good as Time Machine) and I am doing local backups once a week to local USB drives.
I have purchased multiple enclosures from OWC, including a couple of Mercury Pro rack mounts, and found them to be reliable. However I question your choice of using “Scary RAID” (RAID0). Do you really want to pay to download 32TB of data every time you have a problem?
Drives fail, even brand new made for NAS quality drives. I always kept multiple spare drives for our NAS/SAN units and eventually had to use them. IMO, RAID5 is my minimum for a multiple drive unit. And I wouldn’t use that without a local backup of everything (in addition to the cloud BU). Just saying.
You haven’t asked me about the model, but I own a Synology DS718+ (wich is not the current model). I chose that one because of the CPU (I am running a Windows VM on that device [which is a bit slow given the Intel Celeron processor] and PLEX). I have not that much data in comparison to you, so I am fine with 2 bays.
For your use case, a DS420+ might be a good idea, if you change your mind about Synology: DS420+ | Synology Inc.
If you want to go cheaper, a DS 418 would be an alternative. Or a 420j, but the 420j has a really slow CPU.
Clarifying, I wasn’t planning to use any RAID. RAID0 (striping) is indeed much more scary than what I was planning. I’m going for maybe 4 drives, attached in an individually-addressed manner (macOS sees each drive as separate). Most of those drives don’t contain anything even approximating mission-critical data, and I’d be fine downloading it again from the cloud over time. The drive that contains the important stuff would have the important stuff semi-regularly backed up to a local USB drive, with the cloud as a redundancy option.
Am I reading this correctly? You’re running Windows on a Synology?
I’ll look into the Synology options. I’ve heard some horror stories about Synology units failing and all the data being in some weird format that couldn’t be read. That might have been somebody with a dumb config though. If I could pull the drives out and read them in the event of the Synology’s failure, I’d be much more likely to consider moving in that direction.
DS412+, with an external 5-bay extension (DX513 IIRC), giving me nine bays in total.
I also doubled the DS412+'s RAM (porting it to 2Gb), which is an entirely unofficial and unsupported operation, but it works great and you can do it rather easily if you’re even a little careful.
AFAIK, if you don’t use RAID, they’re formatted in ext3 or 4, so that should be readable.
However I’m using mine with SHR (Synology Hybrid Raid) and I already had 3-4 drives fail me during the life of the machine. Each time, I have been warned well in advance, I swapped the faulty disk for a new one, all data was reconstructed in 12-24 hours and it was as good as new.
(And I did some risky things on it like vi’ing into the main config folders to get Crashplan to work on it. This thing is solid as heck.)
Is there anything out there without horror stories about failure or the general end of the world? I do not think so. Everything is fine until it fails eventually. I personally have not had a defective NAS (not with Synology oder QNAP before that). But I have had a lot of defective hard drives. Which was no issue. Because of my NAS and the RAID (RAID1 in my case with a 2-bay device). I pull out the defective HD, put the new one in and everything is fine again, no data loss, no work to do rebuilding the RAID. But yes, you could buy a Synology today and it may fail tomorrow. Or in 10 years. Or whenever.
Nothing is safe without a backup. A RAID will help you with drive failures, but a RAID does not mean not having potential failures of the RAID itself or the device hosting the RAID. And no matter what device you are using for a RAID, it is not trivial to rebuild a RAID if the RAID controller fails. Hence the necessity for a backup. It is the nature of a RAID that you cannot pull out a drive and read it from somewhere else. That is nothing exclusive to Synology. The RAID as a whole contains your data, not a single HD sitting in a NAS. You can only read data from a RAID as long as you access the RAID as a whole.
I am using the Synology Hybrid RAID, too, and I am fine with that. I see it as an evolution to the traditional RAIDs. I have not had any issues with SHR.
I do not see the disadvantages of a RAID. If a hard drive fails within a RAID (1 and above that is), you just need to plug in a new hard drive and you are up and running again. If a hard drive fails outside of a RAID, your data is gone. On a RAID, your data only will be gone when the RAID itself fails. It can happen, but it does not happen very often. Until it happens. That is the reason for a backup even for data on a RAID.
If you do not use the NAS as a RAID but as an enclosure just for plain hard drives, you should be able to read from those drive when you pull them out of the NAS. But honestly, to my mind a NAS does not make much sense then… Something like the OWC device you linked to earlier might be the way to go then.
I may be mistaken, but I was of the impression that a straight-up RAID1 mirror would yield two disks with identical data, with no need for any RAID overhead. In that scenario I’m having a hard time figuring out why those disks wouldn’t be readable outside the NAS, unless the filesystem was proprietary.
Striped data, parity bits, all that? Yeah, failure = scramble. But any idea why a RAID1 mirrored drive would be unreadable outside the NAS?
I have no idea. I have not tried that and I do not plan on doing that. But it very well may be that way.
What I am not 100% sure about is where the actual RAID mirroring stuff happens. (Is it just on the RAID controller’s mainboard or is something also running on the actual hard drives? No idea.). I am no network technician.
I love my Synology. I started with the DS412+ in 2012. It was on a UPS but I think the ethernet got fried during a power outage in 2019 and it wasn’t worth replace the motherboard. I’ve since moved to the DS1819+ which gives me 8 bays and i can purchase an expansion box for more drives. I’m using all 8 bays and run SHR2 which protects me as long as more than 2 drives don’t fail at the same time. I also have 2 external hard drives that the Synology backs up directly.
I’m going for maybe 4 drives, attached in an individually-addressed manner (macOS sees each drive as separate). Most of those drives don’t contain anything even approximating mission-critical data, and I’d be fine downloading it again from the cloud over time. The drive that contains the important stuff would have the important stuff semi-regularly backed up to a local USB drive, with the cloud as a redundancy option.
