Ask us. We’ll tell you plainly which RAID level you have, whether anyone gets the alert when a drive fails, and when your backups were last restored. Call or book a free IT review. No pressure, no jargon.
RAID 10: a dead drive shouldn’t stop your day
Drives fail. That’s normal. The question is whether your office notices.
RAID 10 is the storage layout we recommend for the servers we build, so one failed drive is a quiet swap, not a closed office. It is also not a backup. This page explains why both are true.
What is RAID 10?
RAID 10 (also written RAID 1+0) joins several drives into one. The drives are set up in pairs, and both drives in a pair hold identical data (a mirror). Data is then spread across all the pairs (striping) for speed. It needs at least 4 drives, always an even number, and you can use half of the raw space. Four 4 TB drives give you about 8 TB to work with.
4drives minimum, set up as mirrored pairs
50%of the raw space is usable. The other half is the mirror.
1drive can fail in each mirror pair without losing data
0protection from deletion, corruption or ransomware. RAID is not a backup.
Why it matters
The owner and the person who runs the office care about the same drive for different reasons.
If you own the business
- A dead drive shouldn’t cost a day. In a server with a single drive, a drive failure stops the server, and everyone waits. With RAID 10, the server keeps running while the drive is replaced.
- Downtime has a price. Staff paid to wait, appointments or billable hours lost. RAID removes one of the most common causes.
- It covers one kind of failure only. That’s why we pair it with Hyper-V and tested backups.
If you run the office
- No mystery slowdowns. We monitor the server 24/7, and a failed drive sends email and SMS alerts to our team.
- No downtime to explain. Our servers have hot-swap drive bays, so the failed drive is replaced while people keep working.
- One less thing on you. We order the drive, swap it and watch the rebuild.
How RAID 10 works
Two old ideas, mirroring and striping, working together.
Mirrored pairs
Every drive has a twin holding the same data. If one drive dies, its twin carries on and nobody loses a file.
Striped for speed
Data is spread across every pair, so the work is shared. There’s no parity math on each write, so writes stay fast. Databases, like practice, accounting and ERP software, like that.
Simpler rebuilds
When a new drive goes in, it copies from its twin. A RAID 6 rebuild has to recalculate data from every other drive, which usually takes longer and works the whole array harder.
What RAID 10 survives, and what it doesn’t
It survives
- One failed drive in any mirror pair.
- More than one failed drive, as long as no two are in the same pair. In a 4-drive array, that means up to 2, if they’re in different pairs.
- Normal work carrying on while the failed drive is replaced and the pair rebuilds.
It doesn’t survive
- Both drives in the same pair failing. The whole array is lost and you restore from backup.
- Deleted files, a corrupted database or ransomware. RAID copies them to both twins instantly.
- The server itself failing, a fire, a flood or a theft.
- A power cut in the middle of a write without a battery backup (UPS) and protected controller cache.
Why RAID is not a backup
RAID copies every change to both drives in a pair the moment it happens. That’s the point. It also means a deleted folder, a corrupted database or files encrypted by ransomware are copied just as fast. A backup is a separate copy, kept on separate storage, from a point in time before the problem. We keep 3-2-1 backups with Veeam, an immutable copy locked for 30 days and an encrypted offsite copy in Vancouver, BC. How immutable backups work
RAID 10, RAID 6 or RAID 5?
Every layout trades space, speed and safety. Here’s how we explain it to owners.
- RAID 10: our recommendation. Half the raw space is usable. Fast writes and simpler rebuilds. Survives one failed drive per mirror pair.
- RAID 6: the budget option. Two drives’ worth of space go to parity, so you get more usable space per dollar (at least 4 drives). It survives any two failed drives, but writes are slower and rebuilds take longer.
- RAID 5: we steer away from it for business servers with today’s large drives. It survives only one failure, and a long rebuild leaves no margin if a second drive has trouble.
- RAID 1, a simple two-drive mirror, is common for a small drive that holds only the operating system.
How we set it up
This is how we build the servers we look after.
- Hyper-V, not bare metal, on RAID 10 storage. RAID 6 when space per dollar matters more than speed. We explain the trade-off so the choice is yours. Why Hyper-V
- Matched, business-grade drives of the same model and size, not desktop drives.
- A hardware RAID controller with battery-backed cache, plus a UPS, so a power cut doesn’t lose data that was being written.
- Hot-swap drive bays, so a failed drive comes out and a new one goes in while the server keeps running.
- Monitoring with alerts. A degraded array sends email and SMS alerts to our team, so the failed drive is replaced quickly.
- Room to grow. We never build to a vendor’s bare minimum.
- Backups on separate storage. 3-2-1, Veeam, an immutable copy locked for 30 days, spot checks and a yearly recovery test.
Common mistakes we see
- Calling RAID the backup. It isn’t. See above.
- Ignoring the amber light. A degraded array can run for weeks with no one noticing, with one less safety net in that pair the whole time.
- Desktop drives, or mixed models and sizes, in a server that runs the whole office.
- RAID 5 on large drives to save the cost of one more drive.
- No battery backup (UPS), so a power cut lands in the middle of a write.
- Apps installed straight on the physical server, so the data and the hardware fail together. Hyper-V separates them.
RAID 10 questions
Is RAID a backup?
No. RAID keeps a server running when a drive fails. It doesn’t protect against deletion, corruption, ransomware, theft or a failed server. You still need separate, tested backups.
How many drives does RAID 10 need?
At least 4, always an even number. More pairs add space and speed.
How much space do we lose?
Half. Eight 2 TB drives give 16 TB raw and about 8 TB usable, a little less after formatting.
Can RAID 10 survive two failed drives?
Sometimes. If they’re in different mirror pairs, yes. If both are in the same pair, the array is lost and you restore from backup. That’s why a failed drive gets replaced quickly.
Should we use RAID 6 instead?
If usable space per dollar matters more than write speed and rebuild time, RAID 6 is a reasonable budget option. We’ll show you both and let you choose.
We have no server, only cloud software. Does this matter?
Not for your office. Your cloud provider looks after its own storage. You still want backups of your cloud data, such as Microsoft 365 or Google Workspace.