The storage capacity a CCTV system needs is not determined by the number of cameras alone. Camera resolution, compression type (H.264 / H.265), frame rate (FPS), scene motion, daily recording hours and how many days recordings are kept all directly affect the total storage need. Use the tool below to estimate the approximate disk capacity (GB/TB) required for a home, office, shop or factory; you can also see how many days a disk you already have would record.
How is security camera storage calculated?
Cameras record video by producing a certain amount of data per second (bitrate, Mbps). In the storage calculation this is converted from bits to bytes and multiplied by the recording time. Basic formula: Storage (GB) = Total bitrate (Mbps) × 3600 × daily recording hours × days ÷ 8 ÷ 1000. The most accurate result comes from the real average bitrate shown in the camera or NVR/DVR menu; if unknown, the tool estimates an approximate value from resolution, codec, FPS and motion level.
The values below are approximate starting bitrates for 15 FPS, a medium-motion scene and H.264:
| Resolution | Approx. H.264 bitrate |
|---|---|
| 720p / 1 MP | 1.0 Mbps |
| 1080p / 2 MP | 2.0 Mbps |
| 3 MP | 2.8 Mbps |
| 4 MP | 3.5 Mbps |
| 5 MP | 4.5 Mbps |
| 8 MP / 4K | 6.5 Mbps |
| 12 MP | 9.0 Mbps |
Codec: H.265 works at a lower bitrate than H.264 for a similar quality (about 60% in this tool); smart H.265/H.265+ can drop to ~40% in static scenes. FPS and motion: as frame rate and scene motion rise, data grows; motion-detection recording can reduce storage. Safety margin: the tool adds a default 10% margin to the raw capacity for usable disk size, file system and variable bitrate increases. RAID and backup disks are not included; for critical systems, the real bitrate and a qualified technical assessment should be the basis.
Why does a disk show less capacity than the label says?
You size the system correctly, buy an 8 TB disk, and the NVR menu shows roughly 7.27 TB. This is not a fault. The manufacturer states capacity in decimal units (1 TB = 1,000,000,000,000 bytes) while the operating system counts in binary units (1 TiB = 1,099,511,627,776 bytes).
| Label capacity | Shown by the system | Difference |
|---|---|---|
| 1 TB | ~0.91 TB | 9.1% |
| 4 TB | ~3.64 TB | 9.1% |
| 8 TB | ~7.27 TB | 9.1% |
| 16 TB | ~14.55 TB | 9.1% |
File system overhead and the space the NVR reserves come on top of that. A practical rule: treat about 85% of the label capacity as genuinely usable. The tool above already adds a 10% margin; stepping up one size covers the decimal/binary gap as well.
Worked example: a 16-camera factory system
Scenario: 16 cameras at 4 MP, H.265 codec, 15 FPS, medium-motion scene, 24/7 recording, 30 days retention.
- H.264 starting bitrate for 4 MP: 3.5 Mbps
- About 60% of that with H.265: 3.5 × 0.60 = 2.1 Mbps per camera
- Total: 16 × 2.1 = 33.6 Mbps
- Daily data: 33.6 × 3600 × 24 ÷ 8 ÷ 1000 ≈ 363 GB/day
- 30 days: 363 × 30 ≈ 10,890 GB ≈ 10.9 TB
- With a 10% safety margin: ~12 TB raw requirement
- Once the decimal/binary gap is covered: 2 × 8 TB (16 TB labelled) is a sound choice
The same system built on H.264 would need roughly 18 TB. The codec choice alone changes the cost by about a third — but every camera and the NVR must support H.265.
Why a normal desktop drive should not be used
Surveillance recording is a completely different workload from what a computer disk is used to: continuous, sequential, multi-channel writing. Desktop drives are designed for a few hours a day under variable load.
| Property | Desktop drive | Surveillance drive |
|---|---|---|
| Duty cycle | ~8 hours a day | 24/7 |
| Annual workload rating | ~55 TB/year | ~180 TB/year and above |
| Streaming command set | No | Yes — reduces dropped frames |
| Vibration tolerance | Single-drive use | Multi-bay enclosures |
| Error recovery behaviour | Retries long, can stall the stream | Gives up quickly, keeps the stream alive |
The most common field failure looks like this: the system is installed, runs fine for a few months, then an incident happens and dropped frames are discovered in exactly the critical minutes. The cause is usually a drive that cannot sustain the workload.
Network and PoE budget: the forgotten half of the calculation
Even with the disk capacity sized correctly, the recording arrives over the network. Check two limits:
- Bandwidth: 33.6 Mbps in the example above is nothing for a gigabit network. But a 64-camera 4K system can exceed 400 Mbps, and there the switch backbone and the NVR's incoming bandwidth limit (for example 320 Mbps) become the deciding factor.
- PoE power: A camera typically draws 5–12 W; heated or PTZ cameras go up to 25 W. A 16-port switch with a total PoE budget of, say, 130 W may not feed every camera. Look at the total watt budget, not the port count.
If you want the system to keep recording during an outage, size the combined power of the NVR, switch and cameras with the UPS calculator — backing up only the NVR is useless, because the cameras need power too.
Does RAID remove the need for backup?
No. RAID provides availability, not backup. If one disk fails the system keeps running; but RAID will not recover a deleted recording, an encrypted system or a fire.
- RAID 1: You lose half the capacity; the simplest protection. Sensible for small systems.
- RAID 5: Survives one disk failure. However, with 8 TB and larger drives a rebuild can take days, and a second failure during that window loses the whole array.
- RAID 6: Survives two disk failures. Preferred on systems built with large drives.
Exporting critical footage — an incident recording, for instance — off the system to separate media is a more valuable habit than any RAID level.
The right ways to reduce storage
- Lower the frame rate. For most surveillance work 12–15 FPS is enough. Going from 25 to 15 FPS cuts data by roughly 40% and does not affect incident detection in practice.
- Use motion-based recording — carefully. It saves 90% in an empty warehouse and nothing in a busy production area, where it also risks missing the start of an event. Always enable the pre-record buffer.
- Match resolution to the task. Use 4–8 MP where face or plate recognition is needed, 2 MP for general overview. Making every camera 4K triples the data for limited benefit.
- Turn on the smart codec. H.265+ / Smart codec encodes static backgrounds less often and cuts the bitrate significantly.
How long should recordings be kept?
Technically, as long as your disk holds. But storing footage that contains personal data is a legal matter. Data protection frameworks such as the GDPR do not set a fixed period; under the principles of purpose limitation and proportionality, footage should not be kept longer than the purpose requires. In practice 15–30 days is a common range.
You also need visible signage in monitored areas and a defined retention and deletion policy. This is not legal advice — consult a qualified professional for your own situation. The technical side is simple: decide the retention period first, then size the disk for it. Do it the other way around and the disk capacity ends up dictating your policy.
Frequently asked questions
Where do I find the real bitrate?
The NVR or camera interface usually shows a live value under “Channel status”, “Stream info” or “Bitrate”. If the system has been running for a few days, you can also derive it by dividing the actual disk consumption by the number of days — that is the most reliable method.
Do night recordings take less space?
Usually the opposite. Black-and-white images captured under infrared illumination carry more sensor noise, and noise makes compression harder, so the bitrate can rise. Rain, snow and trees moving in the wind have the same effect.
What happens when the disk is full?
By default the NVR overwrites the oldest recording (loop recording). If that setting is off, recording stops completely — always check it after installation. This is the single setting that causes the most regret in the field.