With SSD storage, what what your thoughts on redundant storage devices (e.g RAID1), is it a waste, a nice to have or a must?
what are your experiences and thoughts on this ?
Appreciate your opinion, as I will probably move from HDD to SSD.
It depends how valuable your data is, what backup strategy you have, and how long you’re prepared to wait to get access to your data when a drive fails.
Personally if/when I migrate my main dataset to SSD, I’ll stick with RAIDZ2/RAID6.
I think it depends on the usage actually. Will there be lots of read and writes? SSDs are quite reliable nowadays.
As someone else has pointed out, it really depends on your use case. Although I personally keep my drives (SSD & HDD) in a redundant RAID configuration as my data is largely mission-critical.
SSD’s are cheap nowadays. They’re also much more reliable than HDD’s. That said, when they fail it’s usually catastrophic failure with total data loss.
If your data that’s important to you, I would still use redundant storage. Especially since the cost is minimal.
The cheaper SSDs can be very failure-prone though. I actually use second-hand spinning disks in my server and I have a whole bunch of spares waiting in the wings. I am still backed up though.
Had a Kingston SATA SSD that I had used for my plex server data (not media), and it ended up dying on me. First drive I had actively fail on me and cause data loss.
It was redundant after that.
That’s interesting because Kingston’s usually have a reputation that proceeds them. I currently have some cheap Chinese knock-off SSD that I’ve put through hell and it keeps plugging away.
Yeah the one I had after it died, I cracked it open.
Was a super cheap PCB that had like 1 or 2 nand packages.
The overall construction of it just felt… cheap, even though at the time it was a higher rated model.
That’s the entire problem with those knock-offs: almost no quality control. So it can be very hit or miss. Bottom line is that if you care about your data, you’ll back it up! 😉
All I gotta say is my media pool has 3 replicas.
2 at home, 1 in Colo.
I never plan on losing data anymore.
Currently my Plex data is split between my old instance at home which has my watch history, and will probably still use when watching 4k.
The new server is on a 5x 12tb z2 pool, lol.
In total I’m probably around 160-170tb raw storage total.
I’ve had SSDs go bad. So if it’s something you really care about I’d approach it the same as hard disks.
At least with spinners you often get enough advanced warning of impending doom via SMART. Critical stuff I am still keeping on spinning disks backed up to tape and a privacy-respecting cloud service.
Ask you a single question: will you break a sweat of you lose your data?
If the answer is anything but a clear, immediate “NO”, go with raid even on SSDs.
(just to add, and clarify - no more than 2 disks)
Maybe as a side note: a mirror raid does not only provide data redundancy but also improves read speeds. For SATA limited SSDs this can be interesting.
SATA SSDs don’t really make sense nowadays anymore. Last time I checked, they cost the same and are actually the exact same thing, except SATA SSDs have a big, mostly empty, plastic case around them and use a slow connector
How, with a couple of hba I can connect connect easily dozens of sata SSDs to a consumer motherboard, while doing the same with nvme drives not exactly as easy or cheap
Just did the math out of curiosity. NVMe averages 4000mb/s (worst case) SATA averages 600mb/s (best case)
It would take about 7 disks to get nearly the same speeds.
Average 4TB NVMe seems to be about 200-250$ 8x cheap as dirt 512MB SSD seems to be about 240$
So if you do not have an NVMe slot on your old mobo but do have 8x spare SATA slots, you could get the same or .5 TB less of space at nearly the same speed, for nearly the same price. You would gain the added benefits of raidZ1 on ZFS, something NVMe on one slot does not give.
This also gets pretty interesting because those could be cheap 1TB disks and you now have 7-8TB of space for around 320$ (depending on raid)
I think it comes down to what kind of motherboard the user has and if they want raid for uptime/disaster recovery.
Sure, except that it is hard to connect several NVMe SSDs to an older mainboard.
Like others said: it depends.
In my opinion the most important thing to have is an off-site backup. If you have achvieved that, it’s only a matter of how long you are willing to wait for the recovery. If your answer is something like “it cannot wait” then you should go with at least raid 1 and consider additional meausres for the event when your whole array fails.
I’m quite disappointed by most comments so far talking about RAID and data loss. That is not what RAID is for at all.
RAID is for uptime/availability. When a drive fails, the system will keep running and working. For companies, that would lose thousands of currency per hour with a downtime, this is super important that the system keeps running. At home, it’s convenience that you can order a new drive and replace without hours of setting up and copying before you can watch the next episode again.
Backups are against data loss. If a single drive fails, a RAID fails or you get some encryption malware or an employee destroys stuff on purpose, then everything is destroyed. It doesn’t matter if it was a single, any RAID, HDD or SSD. You order a new drive, make a new volume and restore the data from your backup.
I have to agree, RAID has only one purpose - keep your data/ storage operating during a disk failure. Does not matter which RAID level or SW. Thank god you mentioned it before.
There can be benefits in addition depending on RAID level and layout, for example read & write speed or more IOP/s than an individual disk (either SSD or HDD). However, the main purpose is still to eliminate a single disk as a single point of failure!
Back to topic - if you have a strong requirement to run your services which (rely) on the SSD storage, even if a disk fails - then SSD Raid yes.
For example.: I have s server running productive instances of Seafile, Gitea, and some minor services. I use them for business. Therefore those services have to be available, even if one disk fails. I cannot wait to restore a backup, wait for a a replacement disk and tell a client, Hey, sorry my server disk failed” (unprofessional)
For protection against data loss - backups: one local on another NAS, one in the cloud. 👌🏼
Well, more specifically it is protecting against a specific form of data loss, which is hardware failure. A good practice if you’re able is to have RAID and an offsite/cloud backup solution.
But if you don’t, don’t feel terrible. When the OVH datacentre had a fire, I lost my server there. But so did a lot of businesses. You’d be amazed at how many had no backup and were demanding that OVH somehow pry their smouldering drives from what remained of the datacentre wing and salvage all the data.
If you care about your data, you want a backup that is off-site. Cloud backup is quite inexpensive these days.
I also thought that way in the beginning, but then disaster recovery is too inconvenient and will take weeks to set everything to your standards, while with raid you just replace the drive and go
Not to mention that “temporary” directory that was supposed to last one week and wasn’t included in the backup script, but then happened to last several months holding important files
Depends on how attached you are on your data, if you have a backup, if you can do without the data created between backup cycles and how long you can wait for the restore to finish.
Everything will fail. For me, everything on single disk is expendable or backupped and can be done without when I loose a day/week of data. Everything else is on raid 1 (hdd) and in a backup schedule (external hdd).
Raid5 (well raidz1) has saved my ass multiple times. Just dont forget to also raid your efi and swap partitions