[MEP Center Engineering]

Starter question — a frequently asked question posted by the MEP Center team to open this forum. Share your own experience below.

A logistics building needs 64 IP cameras at 4 MP, recording continuously for 30 days. The client asks how much NVR storage and what network capacity are needed, and whether 4 MP is enough to identify people at the gates.

• What is the calculation method for bitrate and storage?
• How should RAID and margin be included?
• How do I check that the camera resolution suits the task (detect, recognise, identify)?

[MEP Center Engineering]

Storage = total bitrate × recording time. The bitrate per camera depends on resolution, frame rate, codec and scene activity, so take it from the camera’s calculator for your settings. The resolution check is done with pixel density on the target (EN 62676-4), not with megapixels alone.

Step 1: Bitrate per camera

Typical values for 4 MP with H.265 at 15 fps are 2 to 6 Mbit/s. Busy, night-time or noisy scenes are at the upper end. This example uses 4 Mbit/s, constant bitrate, continuous recording.

Step 2: Total bitrate and storage

1. Total: 64 × 4 = 256 Mbit/s.
2. Per day: 256 / 8 = 32 MB/s × 86 400 s = 2 765 GB, about 2.76 TB/day.
3. 30 days: 2.76 × 30 = 82.9 TB of usable capacity.
4. Add about 10 to 15 % margin (scene changes, file system), giving roughly 92 to 95 TB usable.
5. RAID: with RAID 6, two disks hold parity. For example, 12 × 10 TB disks give 100 TB usable before formatting, which is about 91 TB as the operating system reports it (TB vs TiB). This is just below the target, so choose 12 × 12 TB or two arrays. Always state capacity as usable, after RAID.

Motion-based recording or smart codecs can cut storage by 30 to 70 % in quiet areas, but size continuously recorded cameras without that saving.

Step 3: Network

• Edge switch: 16 cameras × 4 Mbit/s = 64 Mbit/s, so a 1 Gbit/s uplink is plenty.
• Core to NVR: 256 Mbit/s recording plus live viewing and playback streams. Keep sustained load below about 50 % of the link, or use 10 Gbit/s if several viewing clients pull main streams.
• PoE budget per switch: sum the cameras’ maximum PoE class (not typical) plus IR and heater loads in winter.

Step 4: Is 4 MP enough? (EN 62676-4)

The DORI levels are defined by pixels per metre at the target: detect 25, observe 62.5, recognise 125, identify 250 px/m. A 4 MP camera has about 2 560 horizontal pixels, so for identification the field-of-view width at the target must be no more than 2 560 / 250 = 10.2 m. A 12 m wide gate needs two cameras or a higher resolution, and a lens chosen so the 250 px/m is reached at the actual target distance.

Common mistakes

• Using the camera’s minimum bitrate from a brochure for night scenes.
• Quoting raw disk capacity instead of usable capacity after RAID.
• Specifying megapixels without a pixel-density requirement for each camera view.
• Forgetting storage for the retention period required by the client’s policy or local data-protection rules.

For cable routes and racks see Cable Tray Sizing: Fill, Spacing and Derating; for backed-up recording, UPS Sizing: kVA, kW and Battery Autonomy in Practice.

For deeper study: MEP Center online certified courses.

Check the edition of the code that applies to your project.

[Author]

Posts