NVR Storage Capacity: How to Size Recording Storage (2026)

Leo · 2026-11

NVR storage capacity is the total drive space a network video recorder needs to hold footage from every camera over a chosen retention period. The working formula is cameras × bitrate (Mbps) × 86,400 × retention days ÷ 8 ÷ 1,000,000, which yields decimal terabytes. Eight cameras recording at 4 Mbps and kept for 30 days need about 10.4 TB before RAID overhead.

What Is NVR Storage Capacity?

The term NVR storage capacity refers to the amount of usable disk space a recorder can dedicate to video, not the sum of the labels on the installed drives. Formatting, the recorder’s operating system, and RAID redundancy each take a share before the first clip is written. Buyers who treat capacity as a calculated budget instead of a drive-count checkbox avoid the two classic failure modes: an appliance that fills up months early, or shelves of drives that never get used.

woCyber, the export brand of a Beijing-based storage hardware manufacturer with mass-production manufacturing in Shenzhen, publishes capacity ceilings for each of its recorder and NAS models so integrators can run this math against real hardware before issuing a purchase order. The same discipline applies regardless of whose recorder sits on the wall: the specification sheet gives the ceiling, and the site gives the demand.

How Do You Calculate NVR Storage Capacity?

The calculation multiplies stream volume by recording time and converts bits to bytes, in that order. Written out for decimal units:

• Capacity (TB) = cameras × bitrate (Mbps) × 86,400 × retention days ÷ 8 ÷ 1,000,000. The 86,400 term is seconds per day; the 8 converts bits to bytes; the 1,000,000 converts megabytes to decimal terabytes.
• Worked example (illustrative): one camera at 4 Mbps, continuous, for 30 days: 4 × 86,400 × 30 = 10,368,000 megabits; ÷ 8 = 1,296,000 MB; ≈ 1.30 TB per camera per month.
• Scale line: eight cameras under the same assumptions need 8 × 1.30 ≈ 10.4 TB of raw space before any RAID reduction is applied.

The estimate holds only under two conditions. The bitrate input is a planning variable set by resolution, codec, and scene complexity, so it should be confirmed per project rather than copied from a datasheet. And the formula assumes continuous recording; event-triggered recording changes the picture, as the scenario section below shows. NVR storage capacity plans that skip either condition tend to be rewritten within the first year of operation, usually after the array has already filled.

Why Does RAID Reduce Usable Recording Storage?

RAID trades raw capacity for continuity: some drives (or parts of them) are reserved so the array survives a disk failure without losing footage. Recording workloads tolerate this well because video writes are sequential and predictable, and a degraded-but-running array beats a stopped recorder during an incident. The table below shows what remains usable on a four-drive array of 8 TB drives:

For most small and mid-size sites, RAID 5 on a four-bay chassis is the default starting point: it keeps three-quarters of raw capacity while covering a single drive failure. Sites with longer retention windows or sparse on-site service coverage often step up to RAID 6. Both choices belong in the purchase specification, because a recorder delivered without a stated array layout leaves the buyer to discover the usable number after installation.

Bitrate and Retention Move the Estimate the Most

Bitrate scales the result linearly and is the lever buyers most often misjudge, because it is set by equipment choices rather than by the recorder.

Camera count is usually fixed by the floor plan, so the real discussion is between bitrate and retention, and those two levers rarely move together. Doubling retention from 15 to 30 days doubles required capacity at the same bitrate; halving the bitrate does the same thing. Codecs interact with both: H.265 (a video compression standard that targets similar visual quality at roughly half the bitrate of its H.264 predecessor) can pull the bitrate assumption down, but the stream rate a given camera actually produces depends on its resolution, scene complexity, and the camera vendor’s encoder, so integrators quote that number per project rather than from a headline figure. Confirming the bitrate per camera profile before ordering is what keeps the array from filling in month six instead of year three.

Drive Class and Write Workloads

Recording storage also has a duty-cycle dimension that the capacity formula does not capture. Camera streams write to disk around the clock, which is a different profile from desktop storage that sits idle most of the day. Drive makers maintain surveillance-oriented families for exactly this reason: the vendor product page for Seagate’s SkyHawk surveillance drives (queried November 4, 2026) describes them as built around continuous, multi-camera recording workloads. Matching the drive family to the duty profile is a five-minute specification line item; skipping it turns into retries, dropped frames, and warranty conversations later.

Drive class also interacts with RAID rebuilds. A failed drive in a recording array gets rebuilt from parity while the remaining cameras keep writing, so the replacement drive needs to sustain sequential writes without stalling the array. This is one more reason capacity planning and drive specification belong in the same document rather than two procurement steps that never meet. An integrator who quotes the array layout, the drive family, and the rebuild behavior together has usually done this math before; an integrator who quotes only a price per camera has usually not.

A Worked Scenario for a 16-Camera Site

Consider a hypothetical 16-camera vehicle dealership: six cameras inside the showroom and service bays, ten covering the outdoor lot. The owner wants a 30-day retention policy, and the installer assumes a 4 Mbps continuous stream per camera as a conservative planning value.

Continuous recording puts the site at 16 × 1.30 ≈ 20.8 TB per 30 days. A RAID 5 array built from four 8 TB drives provides 24 TB usable, so the continuous plan fits with roughly 3 TB of margin. Motion-triggered recording at an illustrative 50 percent duty cycle would cut the same site to about 10.4 TB, leaving room to add cameras later — but the 50 percent figure is a planning variable to confirm against camera activity on site, and detection tuning on an open parking lot rarely behaves like an indoor corridor. Continuous recording buys predictable coverage at full capacity price; motion recording buys headroom at the cost of tuning effort. The dealer-side cost difference is concrete: at the assumptions above, committing to continuous coverage costs about 10.4 TB more of array space than the motion plan, which approaches the entire monthly footprint of a two-camera site at the same settings. Expansion paths (a larger chassis, bigger drives at replacement time, or a second appliance) deserve pricing before the first purchase order, because retrofitting capacity after an array fills means either a forklift upgrade or a second box with its own management overhead.

Buyer Checks Before Ordering an NVR

Five written confirmations separate a clean deployment from a painful one:

• Confirm the per-camera bitrate assumption with the integrator, in writing, and record the codec behind it.
• Confirm the retention policy in days, and whether recording is continuous, event-triggered, or a mix.
• Confirm the RAID level and the resulting usable capacity — the number that goes into the capacity plan is usable terabytes, not raw drive size.
• Confirm the drive class matches the write workload, using the vendor’s surveillance family documentation as the reference.
• Confirm camera interoperability. ONVIF (the open standard for IP video interoperability, queried November 4, 2026) is what lets a recorder accept mixed camera brands; it is an interface capability, not a certification claim.

Buyers who want recording and private file sharing on one appliance can look at the woCyber S1 security kit, an edge AI NVR plus private cloud NAS in a single desktop unit; its published product page lists retention options of 7, 14, and 30 days, and edge AI detection that classifies people, vehicles, and pets on the device with a published false-alarm rate below 3 percent. Deeper reading on the retention-versus-capacity trade: nvr-storage-planning-for-ip-camera-projects and nas-video-surveillance-retention-storage-sizing cover the same math from the NAS side, and nvr-storage-requirements-for-30-day-retention-in-the-eu adds the European regulatory angle.

The arithmetic here stays simple on purpose. A sixteen-camera site, a 4 Mbps planning bitrate, and a 30-day policy resolve to twenty-odd terabytes and one honest conversation about motion detection. NVR storage capacity turns into a procurement problem only when those three inputs are never written down — at that point every later quote is a guess.