The NVR screen says the retention period is 30 days, but the oldest recording is 12 days old. This difference isn’t a configuration error; when the disk is full, looped recording deletes the oldest data, and the real period is set by the amount of data the cameras produce, not by the setting. The answer to “How many days does this NVR record?” depends on bitrate, recording mode and usable disk capacity. Below are the logic of the calculation, realistic bitrate assumptions, example scenarios and how to verify the calculation on site.
The basic formula
The data a camera produces per day:
Daily data (GB) = bitrate (Mbps) × 86,400 seconds × recording ratio ÷ 8 ÷ 1000
The recording ratio is 1 for continuous recording; for event-based recording it’s the share of time the scene is active. The system total is the sum of all cameras. Required capacity:
Net capacity = total daily data × retention days
Gross capacity = net capacity ÷ RAID efficiency + 10–20% safety margin
The weakest link in the formula is the bitrate value. If the “typical” value from the datasheet or a value measured at noon on installation day is used, the result is optimistic. The right input is the same camera’s average at night and during heavy motion.
What determines bitrate?
- Resolution: Going from 2 MP to 8 MP quadruples the pixel count; the bitrate increase is usually smaller but significant.
- Codec: H.265 targets a lower bitrate than H.264 at the same image quality; the gain varies by scene and encoder (H.265 and H.264 guide).
- Frame rate (FPS): For surveillance recording, 12–15 fps is enough for most scenarios; 25–30 fps increases data.
- Scene complexity: Tree leaves, rain, crowds and night noise raise bitrate. The same camera can produce 2 Mbps by day and 5–6 Mbps at night.
- Bitrate mode and GOP: VBR varies with the scene; CBR is constant but wastes either quality or capacity. GOP length and smart codec features reduce data in static scenes.
A common situation: installation is done in summer, measurements are taken by day and the disk is chosen accordingly. In November the nights get longer, noise under IR increases and the retention period quietly shrinks. Nobody touched the settings. The scene changed.
A warning about smart codec and long GOP settings: these modes are camera- and manufacturer-specific. They may not always be supported in third-party VMS products, or they may cause frame skipping when rewinding. Before enabling them, check the release notes of the VMS in use and test playback.
Example calculations
The values below are average estimates for illustration; real values must be measured with the camera model and scene. The continuous recording rows are approximate video volumes calculated assuming 24-hour uninterrupted recording and a constant average data rate: 16 cameras × 1.5 Mb/s × 30 days ≈ 7.78 TB, 32 cameras × 3 Mb/s × 30 days ≈ 31.10 TB (TB = 10¹² bytes). Usable capacity after RAID, audio and extra streams, space reserved by the device and growth headroom aren’t included in this volume; each is calculated separately.
| Scenario | Assumption | Daily data | 30 days net |
|---|---|---|---|
| 16-camera office | 2 MP, H.265, 15 fps, avg. 1.5 Mbps, continuous | 16 × 16.2 GB ≈ 259 GB | ≈ 7.8 TB |
| 32-camera warehouse | 4 MP, H.265, 15 fps, avg. 3 Mbps, continuous | 32 × 32.4 GB ≈ 1.04 TB | ≈ 31 TB |
| 64-camera factory | 4 MP, H.265, 12 fps, avg. 2.5 Mbps, 60% event recording | 64 × 16.2 GB ≈ 1.04 TB | ≈ 31 TB |
| 8 outdoor cameras | 8 MP, H.265, 15 fps, avg. 6 Mbps, continuous | 8 × 64.8 GB ≈ 518 GB | ≈ 15.6 TB |
Example calculation step: 3 Mbps × 86,400 s = 259,200 Mb; ÷ 8 = 32,400 MB ≈ 32.4 GB/day.
Note the factory row. Event-based recording on 64 cameras needs the same capacity as continuous recording on 32 cameras. Event recording isn’t always a 50% saving; in an area with shift work and constant forklift traffic, the recording ratio approaches 1. Pre- and post-recording times also add a few seconds to every event.
RAID and usable space
In the 32-camera warehouse example, if 8 × 6 TB disks are configured with RAID 6 for 31 TB of net capacity, raw usable space is 6 disks × 6 TB = 36 TB. Disk manufacturers label TB as 1012 bytes; operating systems mostly show TiB (240), so the visible space is ≈ 32.7 TiB. Compare in the same unit: 31.10 TB of video volume is ≈ 28.3 TiB. Once file system overhead, space reserved by the device and growth headroom are added, the 30-day goal is borderline; in that case either 8 TB disks or recording profile optimization is considered. For detailed capacity planning, see our storage planning service.
RAID choice isn’t only about capacity. Rebuild time depends on capacity, effective speed, recording load and storage architecture; during this time the disks are under heavy read load. RAID provides tolerance to disk failure; it doesn’t replace an independent recording backup. In a large array built with RAID 5, a second disk error during rebuild loses the whole array. That’s why RAID 6 and a hot spare are the common choice in multi-disk recording systems. Disks are also chosen from surveillance or enterprise-class models built for 24/7 write workloads; desktop disks aren’t designed for this workload.
The NVR’s incoming bandwidth limit
As important as disk capacity is the total bandwidth the NVR can accept at once (for example 160 or 320 Mbps). 32 cameras × 3 Mbps = 96 Mbps of recording streams, plus remote viewing and playback load. Devices running near the limit can produce frame loss and recording gaps; leaving at least 30% headroom in the design is good practice.
The incoming bandwidth value on the datasheet is often for recording only; several operators playing back or exporting at the same time creates read load on the disks. If a system that runs exports during the busy night-time event hours develops recording gaps, this total load is the first place to look.
Whose decision is the retention period?
The technical calculation answers “how many days can we keep?”; “how many days should we keep?” is an organizational and legal decision. The Turkish Personal Data Protection Authority’s announcement on security camera use in workplaces emphasizes keeping recordings only as long as necessary and preferring the shortest possible period. No single number of days is used for general camera applications. Any sector-specific obligation, the purpose of recording and the necessary period are assessed together. The period should be set taking into account how long it takes for incidents to be noticed and internal review processes, and applied in the system as the maximum retention (KVKK and retention guide).
In practice, two settings are used together: disk capacity that meets the agreed period, and that period defined as the maximum retention in the NVR/VMS. If you rely only on looped recording, as disks grow recordings are kept beyond the period in the policy.
How do you verify the calculation on site?
- After installation, record actual bitrate values per camera for at least a week (day and night separately).
- Check weekly by opening and playing the oldest recording in the NVR/VMS; the oldest recording that can actually be played back shows the real retention period.
- At acceptance of a new installation, if the target period hasn’t elapsed yet, record the calculation and the measured data rate at initial acceptance; do the actual retention verification separately once the period has elapsed.
- Monitor the disk fill curve; update the calculation when new cameras are added.
- Test that disk and RAID warnings arrive as notifications (NVR installation and management).
This measurement also gives you the list of cameras producing the most bitrate. Usually a few cameras produce most of the total: an outdoor camera looking at a tree swaying in the wind, a corridor camera whose IR reflects off a nearby wall. Fixing these cameras’ angle, night settings or frame rate is the first step to try before adding disks.
Frequently asked questions
The NVR setting says 30 days, but recordings are deleted after 12 days. Why?
The period in the setting is a maximum target; when the disk is full, looped recording deletes the oldest data. The real period is determined by the total bitrate and disk capacity. Night noise and heavy motion increase bitrate and shorten the period.
How much does motion-triggered recording save?
It depends on activity in the scene. In an empty corridor at night it saves a lot, while on a dock with constant traffic the difference is small. Pre- and post-recording times must also be added.
How much does RAID reduce capacity?
RAID 5 uses one disk’s capacity for parity, RAID 6 two; RAID 1/10 uses half the capacity. Also, because of the file system and the TB/TiB difference, usable space is below the label value.
If capacity isn’t enough, optimizing the recording profile first (FPS, smart codec, night settings) and then adding disks is usually the most economical path. For failed or full disks, see our disk and capacity upgrade service.