All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
multi-drive dedicated servers so hard to find. Is RAID still necessary?
I’m planning to migrate my workloads from a VPS to a dedicated server. Since most VPS hosts use RAID on the hypervisor to provide some storage redundancy (I know, RAID is not a backup!), I’m naturally leaning towards a RAID 1 setup for my first dedicated server.
However, I’ve noticed that most instant-setup/pre-configured dedicated servers nowadays come with a single massive NVMe drive (like 3.84TB or 7.68TB) rather than two smaller drives.
On the other hand, fully customizable servers are way more expensive. For instance, on an 8-core Ryzen / 64GB RAM / 1TB NVMe base model, adding an extra 32GB of RAM and a second 1TB NVMe drive costs almost as much as the monthly base price itself! (I know, that's a budget issue on my end, haha 
).
This got me thinking: Does the market nowadays just prefer single-drive configurations?
Also, what are the failure rates of modern NVMe drives like? Since I already have solid off-site backups in place, do I really still need RAID 1, or is running a single drive perfectly fine for most use cases today?
Would love to hear your thoughts and experiences!
Comments
Depends on how much you value your data/uptime i guess. If the drive fails it will depend on your backup strategy and how fast you can pull off a full restore vs. shutdown, switch drive, reboot (assuming you can't just hotswap).
Drive failures have never been that much of a regular thing (maybe outside of getting old drives/heavy usage) but when they happen it tends to be the most inconvenient timing possible and i'd rather arrange for an orderly replacement than be down right now and scramble to get back up (replacing drive + full restore from backup) at 3:23am.
The key thing is that RAID 1 and your off-site backups solve different problems. Backups protect your data; RAID 1 protects your uptime. Since you already have solid, tested off-site backups, the real question is: how much does an unplanned outage plus a restore window cost you?
If a few hours of downtime while you re-provision and restore is acceptable (personal projects, dev boxes, anything non-customer-facing), a single NVMe is genuinely fine and lots of people run exactly that way. One caveat on failure mode: NVMe annualized failure rates are low, but when they go they tend to go abruptly and completely, rather than throwing the gradual SMART warnings you'd get from a spinning disk. So plan for "sudden total loss until restore," not "degraded and limping."
RAID 1 earns its keep when staying online through a single drive failure matters — you keep serving while you arrange a hot-swap/rebuild instead of restoring from backup. For production stuff with users, that's usually worth the second drive.
Two things I'd prioritize over RAID either way: (1) ask the provider whether the NVMe is consumer or enterprise/datacenter-grade — endurance and sustained-write behaviour differ a lot, and that matters more than RAID for many workloads; (2) actually test-restore your backups end to end and time it, so "I have backups" becomes "I can be back up in X minutes." If that number is acceptable and the dual-drive option nearly doubles your cost, single enterprise NVMe + tested restores is a perfectly defensible budget call.
If you have a good backup strategy and 24-48 hr downtime is not an issue for u, then u can go with a singe disk.
However, I can't imagine a usecase where u need 4-8 TB NVME SSD AND are willing to accept 24-48 hr downtime.
Thats why i like HDD servers like OVH 4x2 TB < 20$ / Month.
They read / write with around 100MB/s thats the network speed 1 gig, and cheap!
We normally do 2-drive configurations as a default,
very few (if almost any) would have 1, but just to try to lower down the price anyway, because of the absurd condition of the current hardware market.
Thank you for your analysis. I'm fortunate that I've never encountered a drive failure.
Because the entire environment is manually set up, restoring from backups and troubleshooting could potentially take a whole day.
Considering my backup strategy is every 24 hours (planned to upgrade to more frequent), this would still result in some data loss, which could be quite significant. Therefore, I've opted for RAID 1.
I prefer to store all static content on SSDs. Currently, the total data volume is less than 1.5TB, so the price is within an acceptable range.
If the data volume becomes significantly larger in the future, I will definitely have to choose an HDD RAID array.
OVH is indeed excellent. Unfortunately, I need a US-based hosting provider.
Otherwise, both OVH and Hetzner would be excellent choices.
RAID 1 for that profile makes sense, and the reason lines up exactly with your worry about a full-day restore: RAID 1 is really about shrinking the restore window, not about data safety. A single disk dies, you stay online and rebuild from the mirror — no restore-from-backup, no day of downtime. That's a different job from your 24h backups, which protect against the things RAID can't see: an fs-level corruption, a bad deploy, an accidental rm, ransomware. Both copies in a RAID 1 happily mirror all of those instantly.
So I'd frame it as two separate dials rather than RAID-vs-backups. For <1.5 TB of mostly-static content, the cheapest way to close your data-loss gap isn't a bigger array — it's tightening the backup interval. rsync/restic snapshots to a second cheap box hourly (or even every 15 min for the small delta on static data) costs almost nothing and turns "significant data loss" into "a few minutes". Then RAID 1 handles uptime, backups handle integrity, and neither is doing the other's job.
OVH has a US branch, OVHCloud US, that is a completely separate US entity with their own US-based team.
But yeah, I'd never recommend running single drive dedicated servers, and I don't think any server provider should be selling dedicated servers with single drives in 2026; but, with the NVMe prices now like 300-600% what they were a year or two ago, it's rough.
Wow, I found it! But US servers are much more expensive than those in other regions, so it doesn't seem like a good deal.
You should also consider OVH Canadian DC in BHS, there are good connection to US.