Surveillance storage is a different problem from desktop storage. A single 4K camera can write 15-20GB per hour, continuously, for months at a time. Multiply that by 8 or 16 cameras and you’re looking at a workload that spends most hard drives and a lot of SSDs faster than people expect. The right choice depends less on speed and more on how the drive handles sustained, repetitive writing.
Why surveillance workloads are unusual
Most consumer storage is bought for mixed read/write use: loading games, saving documents, occasional large file transfers. A DVR or NVR does almost nothing but write, in a loop, 24/7. Old footage gets overwritten as new footage comes in. That’s a workload defined by sustained sequential writes and high total bytes written (TBW), not by random IOPS or boot times.
This matters because consumer SSDs are rated for maybe 150-600TB of total writes over their life. A 16-camera system writing 300GB/day blows through 300TB in under three years. A surveillance-grade HDD, by contrast, is rated for continuous operation and millions of write cycles, because spinning platters don’t wear out from writes the way NAND does.
SSD vs HDD for DVR/NVR storage
| Factor | HDD (surveillance-rated) | SSD (consumer/endurance) |
|---|---|---|
| Cost per TB | $18-25 | $45-70 |
| Write endurance | Not wear-limited; rated for 24/7 duty cycles | 150-600+ TBW depending on model |
| Typical failure mode | Gradual (bad sectors, reallocation) or mechanical seizure | Sudden drop to read-only once NAND wears out |
| Noise/vibration tolerance | Can degrade in multi-bay rack vibration | No moving parts, unaffected |
| Power draw | 5-8W active | 2-4W active |
| Best use case | Primary DVR/NVR storage, large capacity archives | Edge devices, short-retention buffers, boot drive in NVR appliance |
For almost all home and small-business setups, a surveillance-rated HDD is still the right default. They’re built for exactly this workload, cost a third as much per terabyte, and their failure mode is usually gradual rather than instant. SMART attributes like reallocated sector count give you warning before total failure. If you’re shopping for one, look specifically at surveillance-rated hard drives rather than generic desktop or even NAS drives — they’re firmware-tuned for continuous sequential writes and lower vibration sensitivity in multi-bay enclosures.
When an SSD actually makes sense
SSDs aren’t wrong for surveillance, they’re just usually overkill or mismatched to the workload. They make sense in a few specific cases:
Edge recording on a single camera or small 2-4 camera setup with short retention (a few days of loop recording) won’t generate enough total writes to matter — you’d need years to hit endurance limits. Here cost difference is small enough that an SSD’s silence and shock resistance (useful in vehicles, kiosks, or industrial settings) can justify the premium.
NVR appliance boot/OS drives benefit from an SSD too, since the OS itself isn’t the thing taking the sustained write hammering — the camera footage is. Many commercial NVRs already separate OS storage from footage storage for this reason.
High-bitrate forensic or AI-analytics setups that need fast random access for simultaneous multi-stream playback and motion analysis can benefit from SSD read speed. If you’re building something like this, an endurance-focused option from the high-endurance SSD category (rated in DWPD, drive writes per day, rather than consumer TBW) is worth the extra cost, because consumer SSDs will genuinely die early here.
RAID doesn’t fix the wrong drive choice
A common mistake is assuming RAID 1 or RAID 5 solves endurance or reliability problems. It doesn’t — it just means two drives wear out on the same schedule instead of one failing alone. If you’re using consumer SSDs in a mirrored NVR setup, both drives are absorbing identical write patterns and will likely approach end-of-life within weeks of each other. RAID protects against a single drive failure event, not against simultaneous wear-out. For long-retention systems, stick with HDDs in RAID 1 or RAID 5, and treat RAID as protection against a dead drive, not a substitute for choosing the right drive type to begin with.
Sizing your storage correctly
Work backward from retention days, not forward from drive size. A single 1080p camera at a typical 4Mbps bitrate uses about 40GB/day; a 4K camera at 15Mbps uses roughly 150GB/day. For 8 cameras at 4K with 30-day retention, that’s about 36TB — meaning you want two 20TB drives in RAID 1, or four smaller drives in RAID 5 for both capacity and redundancy. Undersizing capacity is the single most common mistake in DIY camera systems; people buy based on camera count and forget retention length entirely.
What to actually buy
If you’re setting up a standard home or small business NVR with more than a couple cameras and more than a few days of retention, buy surveillance-rated HDDs. They’re cheaper, built for the exact write pattern you’re generating, and fail predictably. Save SSDs for the NVR’s own OS drive, short-retention edge devices, or specialized high-throughput analytics boxes where random read speed genuinely matters. Don’t let an SSD’s reputation for speed and reliability in desktop use convince you it’s automatically the safer choice here — in this specific workload, it’s often the one more likely to die first.
