FULLCOLORCAM

CCTV Storage Guide: Recording Calculator, Hard Drive Sizing and Retention (2026)

RCRay Chan·2026-08-27·15 min read
CCTV storage calculator and hard drive sizing
Table of Contents

Most buyers pick a CCTV hard drive by price and discover weeks later that their recorder overwrites footage every four days. Storage looks like the easiest part of a camera system, yet it is where most installations fail: the cameras are fine, the network is fine, and the hard drive size was simply guessed. The number that decides everything is GB per day, and it swings wildly with resolution, frame rate and compression. A 16-camera site at 4MP with H.264 needs roughly 9-12 TB to keep 30 days of recording; the same cameras on H.265+ fit in 4-6 TB. Same cameras, half the hardware bill — that gap is what a proper CCTV storage plan buys you.

This guide gives you a working storage calculator, a GB-per-day table for 2MP/4MP/5MP/8MP across H.264, H.265 and H.265+, the math for converting capacity into recording days, an honest look at surveillance hard drives, RAID levels, cloud storage vs local recording, and a checklist you can hand to any supplier. By the end you will be able to spec the video storage for a site from a camera list and a retention target, without guesswork.

Keep reading for more!

The Snapshot

  • 1 Mbps of average bitrate writes about 10.8 GB per day — the single conversion factor behind every CCTV storage calculator (1 Mbps × 3,600 seconds × 24 hours ÷ 8 = 10.8 GB).
  • A 4MP camera records roughly 18-25 GB/day with H.264, 9-13 GB/day with H.265 and 6-10 GB/day with H.265+ on a typical mixed scene at 15 fps — the codec choice moves storage more than the resolution choice.
  • Sixteen 4MP cameras on H.265+ keeping 30 days need about a 4-6 TB recorder; the same cameras on H.264 need 9-12 TB. Compression is the cheapest storage upgrade you can buy.
  • Surveillance-rated drives are built for 24/7 recording (around 180 TB/year workload); desktop drives are rated for about 8 hours a day and their error-recovery routines can stall a recorder mid-write.
  • RAID 5 on four drives gives (N-1)/N usable capacity — 30 TB usable from 4×10 TB — and protects a week-plus of footage against a single drive failure, which is why 16+ camera sites use it.

What Drives CCTV Storage: Resolution, Frame Rate, Compression

Every byte on a recorder comes from the same chain: a camera captures pixels, encodes them into a stream, and the recorder writes that stream to disk. Four variables decide how many bytes per second that is — resolution, frame rate, compression and scene activity. The storage per day is roughly pixels per frame × frames per second ÷ codec efficiency × activity factor, and each variable multiplies or divides the bill.

Resolution is the pixel count per frame. A 2MP camera produces 1920×1080, a 4MP camera 2560×1440, a 5MP camera 2560×1920 and an 8MP camera 3840×2160. The jump from 2MP to 8MP is four times the pixels, and at the same quality level the bitrate climbs in roughly the same proportion, which is why an 8MP site needs about four times the storage of a 2MP site with the same codec, frame rate and scene. The resolution-to-storage relationship is laid out numerically in our megapixel IP camera guide, and the full camera selection process in our security camera guide.

Frame rate is how many frames per second are encoded, typically 15 or 25 fps for CCTV, 30 fps for smoother playback. Doubling the frame rate does not double storage, because consecutive frames share most of their content and the encoder only stores the differences; in practice 30 fps costs about 30-50% more than 15 fps on the same camera.

Compression is the biggest lever of all. H.264 was the CCTV baseline for a decade; H.265 (HEVC) roughly halves the bitrate at the same quality, and the enhanced H.265+ mode adds scene-aware bitrate control that drops the rate further when nothing is moving. A camera that records 18-25 GB/day on H.264 falls to 6-10 GB/day on H.265+ — a 60-70% reduction. The technical comparison between the two codec generations is covered in our H.265 vs HEVC guide, and the practical storage savings in our H.265 vs H.264 guide.

Scene activity is the variable most buyers forget. A camera pointed at an empty corridor at night encodes almost nothing; the same camera over a busy loading dock encodes constantly. H.265+ and motion-based recording modes exploit this automatically, which is why two identical cameras on the same site can write very different amounts per day. Any capacity plan that ignores activity will either over-buy or under-provision.

The CCTV Storage Calculator: Bitrate to GB per Day

The core conversion of any CCTV storage calculator is bitrate to bytes. One megabit per second sustained (the average stream size) works out to 10.8 GB per day: 1 Mbps × 3,600 seconds per hour × 24 hours per day = 86,400 megabits per day, divided by 8 = 10,800 MB, or 10.8 GB. To estimate a camera's daily footprint, take its average bitrate in Mbps and multiply by 10.8.

Here is the calculation with a real camera. A 4MP camera on H.265+ is typically configured with a 2-4 Mbps bitrate cap. At the 2 Mbps cap, constant-bitrate recording writes up to 21.6 GB/day — the ceiling. But on a typical mixed scene, H.265+ scene-aware mode only writes when the image changes, and the sustained average lands around 0.6-1.0 Mbps, which produces the 6-10 GB/day seen in practice. The configured bitrate cap gives you the worst case; the scene-aware average gives you the real number. Quoting peak bitrate for capacity planning over-provisions storage by three to four times.

Once you have per-camera GB/day, the days calculation is simple division: recording days = usable capacity in GB ÷ (number of cameras × GB per day per camera). A site with 8 cameras at 10 GB/day writes 80 GB/day; an 8 TB recorder (7.3 TB usable in practice) holds roughly 90 days. A 16-camera site at the same 10 GB/day writes 160 GB/day, so a 4 TB recorder holds about 25 days and a 6 TB recorder about 37 days.

Always add headroom to the final number — 20-30% is the working rule among integrators. Drives fill, files fragment, firmware upgrades double-write segments, and a busy holiday season can push a retail site's activity factor far above the estimate. A plan sized to exactly 30 days will overwrite at day 28.

Security camera mounted under the eave feeding an NVR

GB per Day by Resolution and Codec: 2MP to 8MP

The table below is the reference most buyers actually need: per-camera GB per day for every common CCTV resolution across the three codec generations, at 15 fps on a typical mixed scene with about 30% activity. Figures are sustained averages, not peak bitrates.

ResolutionPixelsH.264 (GB/day)H.265 (GB/day)H.265+ (GB/day)
2MP1920×10809-136-84-6
4MP2560×144018-259-136-10
5MP2560×192020-2812-1610-14
8MP3840×216032-4818-2616-24

← swipe to scroll →

Three patterns jump out of the table. First, moving from 2MP to 4MP roughly doubles storage at every codec level — expect the same on the drive bill. Second, 5MP costs only modestly more than 4MP (its 4:3 frame holds 2560×1920 pixels), which is why it remains the sweet spot for doorways and tall scenes. Third, the codec choice moves the needle more than the resolution choice: a 4MP camera on H.264 (18-25 GB/day) costs more to store than an 8MP camera on H.265+ (16-24 GB/day).

The 4MP row matches the figures published in our megapixel guide, and the other rows follow the same measured relationship: H.264 runs roughly double H.265+, with plain H.265 landing between the two. All values are typical: a camera pointed at a busy street at 30 fps sits at the top of its range, while an empty warehouse at 10 fps can sit below the bottom.

How Many Recording Days Will a Hard Drive Hold?

Recording days is the question every buyer actually asks, and the answer is always the same division: capacity divided by daily write volume. The hard drive size you need equals cameras × GB/day × target days, plus 20-30% headroom. The matrix below shows what a site of 4 to 32 cameras keeps on a given capacity, using the 4MP H.265+ midpoint of roughly 10 GB/day per camera.

Cameras (4MP H.265+)7 days30 days60 days90 days
4 cameras≈0.3 TB≈1.2 TB≈2.4 TB≈3.6 TB
8 cameras≈0.6 TB≈2.4 TB≈4.8 TB≈7.2 TB
16 cameras≈1.1 TB≈4.8 TB≈9.6 TB≈14.4 TB
32 cameras≈2.2 TB≈9.6 TB≈19.2 TB≈28.8 TB

← swipe to scroll →

Work through a common example. A small business with 8 cameras at 4MP H.265+ wants 30 days: 8 × 10 GB × 30 = 2.4 TB, so a 4 TB drive (or two 2 TB drives in RAID 1) is the right purchase. A 16-camera warehouse wants 60 days: 16 × 10 × 60 = 9.6 TB, which means a 4-bay recorder with 4×4 TB in RAID 5 (about 12 TB usable) or 2×6 TB in RAID 1.

What retention target should you plan for? Thirty days is the working baseline for most businesses; 60-90 days is common for retail, schools and logistics; 90+ days is standard where regulators or insurers ask for it (point-of-sale and cash-handling areas routinely require 90 days). Check your insurance policy wording before you size the drives: many policies now state a minimum retention period, and the cheapest way to fail an audit is a recorder that overwrites at day 25.

Frame Rate and Recording Modes: Continuous vs Motion

Frame rate and recording mode are the two settings that let you tune storage without touching resolution. Dropping a camera from 25 fps to 15 fps typically saves 25-35% of its daily footprint; dropping to 10 fps saves more, at the cost of choppier motion for fast-moving subjects. For a camera watching a gate or a corridor, 10-15 fps is usually plenty; for a camera watching a cash desk where gestures matter, keep 25-30 fps.

The larger lever is the recording mode. Continuous 24/7 recording writes the full daily footprint every day. Motion-only recording writes only when the camera detects activity — on a typical site with 30% activity, that cuts storage by 50-70%. Event-based sub-stream recording (a low-resolution stream triggered by alarms) can drop a low-traffic site's total to a tenth of continuous. The practical pattern is per-channel: continuous at tills and entrances, motion-only on yards and corridors, sub-stream events on perimeter cameras.

The trade-off is evidence risk. Motion-only recording depends on the motion detection firing: a camera aimed at a dim area with detection zones drawn too small can miss the moment you need. The safe default is continuous recording on channels that matter, motion-only on low-value coverage, and a retention policy long enough that the loop overwrite never touches a working day before you export it.

CCTV Hard Drives: Surveillance vs Desktop vs Enterprise

Not all drives are built for CCTV, and the difference shows up in the second year of 24/7 operation. A camera system writes to disk continuously, in a streaming pattern, with heavy bursts at night when activity spikes, a workload that desktop drives were never rated for. A purpose-built cctv hard drive is designed for exactly this duty cycle.

1. Surveillance-rated drives (WD Purple-class)

Surveillance drives are rated for 24/7 operation with a workload of roughly 180 TB per year, use ATA streaming commands so the recorder's writes are prioritized, and, critically, use short error-recovery controls (TLER, about 7-15 seconds) so a read hiccup does not stall the recorder or drop the drive out of a RAID array. This is the default choice for any NVR or DVR bay.

2. Desktop drives

Desktop drives are engineered for about 8 hours of use per day. In a recorder they run hot, wear faster, and their aggressive error-recovery routine can pause the drive for up to two minutes trying to fix a bad sector, during which the recorder keeps writing to a drive that is not accepting data. In a RAID array that pause triggers a drive removal. They work in a pinch for small systems with short retention, but they are a false economy on any site that matters.

3. Enterprise and NAS drives

Enterprise drives carry the highest workload ratings (around 550 TB per year) and the same short error recovery as surveillance drives. They are built for servers and high-bay storage servers running VMS software. NAS drives sit between the two — 24/7 rated with RAID-friendly error recovery, a solid choice for a NAS-based recording setup. For a dedicated recorder, surveillance drives give the best cost-to-duty balance; for a server running video management software, enterprise or NAS drives fit better.

4. CMR vs SMR: pick CMR

Within any of these families, choose CMR (conventional magnetic recording) over SMR (shingled). SMR drives rewrite overlapping tracks, and their sustained write speed collapses under continuous recording load. Every major drive maker marks the recording technology on the label; if it is not marked, ask. CMR is the safe buy for every camera workload.

Practical capacities: 4 TB and 6 TB drives are the workhorses of 1-2 bay recorders; 8-16 TB drives fill 4-8 bay units; 20 TB drives exist for high-density storage servers. All CCTV recorders use 3.5-inch SATA drives; confirm the bay size and max drive size supported before buying, because many recorders cap per-bay capacity at 8 TB or 16 TB regardless of what the drive can hold.

RAID for NVR and DVR: Mirror, Parity and When to Bother

RAID exists for one reason: a single drive failure should not erase your history. A 30-day retention policy is worthless if drive number two dies on day 29. RAID levels trade capacity for protection, and the right level depends on how many bays your recorder has and how much footage you can afford to lose.

  • RAID 1 (mirror): two drives, identical copies, 50% usable capacity. The default for 2-bay recorders — 2×4 TB gives 4 TB usable but survives one drive dying.
  • RAID 5 (single parity): three or more drives, one parity drive's worth of overhead, (N-1)/N usable. Four 10 TB drives give 30 TB usable. The standard for 4-8 bay recorders.
  • RAID 6 (dual parity): four or more drives, (N-2)/N usable, survives two simultaneous failures. Worth it on 8-bay units with 16+ TB drives where rebuilds take days.
  • RAID 10 (mirror + stripe): four or more drives, 50% usable, fastest rebuilds and best performance — the choice for storage servers, rarely needed inside a recorder.

The rebuild reality matters more than the marketing. Rebuilding a 10 TB drive in RAID 5 takes roughly 12-24 hours under load, and during the rebuild the array is running without protection; a second failure loses everything. RAID 6 exists precisely for that window. On a 2-bay recorder, RAID 1 is the only sane choice; on 4 bays, RAID 5; on 8 bays with large drives, RAID 6.

Two practical notes for integrators. First, RAID needs drives with short error recovery, which is why the CMR surveillance drives from the previous section are the right partner; desktop drives get dropped from arrays. Second, not every recorder supports RAID: many 1-bay and some 4-bay units do not, so check the spec sheet, and regardless of RAID, pair the recorder with a UPS so a power cut cannot corrupt the parity set mid-write.

Technician configuring a network video recorder system

Cloud Storage vs Local: The Bandwidth Reality

Cloud storage sounds like the modern answer — no drives, no RAID, no local hardware. The reality is that cloud video storage is a bandwidth game, and continuous multi-camera recording usually loses it. The comparison below is what a buyer should weigh.

DimensionLocal NVR / DVR storageCloud storage
Cost modelOne-time hardware purchaseMonthly per camera or per TB, ongoing
Bandwidth neededNone for recordingSustained upload per camera (2-4 Mbps each)
RetentionLimited by drive capacity, 30-90 days typicalUnlimited in theory, priced per TB
Risk profileLocal theft or fire can destroy footageOff-site copy survives local events
PlaybackInstant, on the local networkDepends on upload speed and latency
Best fitContinuous 24/7 multi-camera sitesEvent clips, alerts, single-camera sites

← swipe to scroll →

Do the upload math before signing up. Sixteen cameras each streaming a 2 Mbps main stream need 32 Mbps of sustained upload, and most business broadband plans provide 10-30 Mbps upload, and that is before anything else uses the line. The recording simply cannot keep up, and the cloud provider will either throttle, buffer or bill you for a faster line.

History upload is even worse. Eight cameras at 10 GB/day generate 300 GB per month; uploading 30 days of history (about 2.4 TB) over a 20 Mbps line takes roughly 11 days of continuous upload, meaning the cloud copy is always a month behind. The honest architecture is hybrid: record continuously on a local recorder, and push events or short clips to the cloud for off-site backup and alerts. That gives you the evidence-grade local record plus an off-site copy of the moments that matter, for a fraction of the bandwidth.

Cloud-only makes sense in narrow cases: a single camera at a vacation property, a temporary construction site without a recorder, or sites where the recorder itself is at theft risk. Budget on typical pricing of $3-10 per camera per month for cloud video: a 16-camera site is $576-1,920 per year that a local recorder writes off in year one.

NVR and DVR Storage Limits: What the Recorder Allows

Before you buy drives, read the recorder's storage spec, because NVR storage capacity is bounded by four hardware numbers: bay count, max drive size per bay, total recording bandwidth and codec support. A 4-bay NVR that caps each bay at 8 TB tops out at 32 TB regardless of the drives you slot in.

Bay count runs from 1 to 8 on mainstream recorders. One bay suits a small shop with short retention; 2 bays allow RAID 1; 4 bays are the B2B workhorse (RAID 5 with headroom); 8 bays serve 32-64 channel sites and can hold months of footage. Max drive size per bay is the number buyers miss: many recorders cap at 8 TB or 16 TB per bay, so confirm it against the drive you plan to buy. Total recording bandwidth (for example, 256 Mbps in) sets how much simultaneous 8MP recording the unit can actually accept, and H.265+ decode support decides whether the camera's encoding savings survive on disk.

DVR storage works exactly the same way for analog and HD-over-coax (AHD/TVI/CVI) cameras: the GB-per-day math, drive types and RAID guidance above apply unchanged, with the recorder doing the encoding instead of the camera. If you run coax cameras, see our DVR and AHD guide and DVR products; for IP camera systems the full recorder buying manual is our NVR guide, and the complete hardware is on the NVR product page.

Two expansion paths exist when the bays fill. eSATA ports let a recorder attach external drive enclosures, and a network storage server (NAS or a dedicated storage box) can take recording off the recorder entirely for very large sites. Kit buyers should confirm what is bundled: many CCTV kits ship with a drive pre-installed, while bare recorders do not; add the drive cost to the quote before comparing prices.

Storage Purchasing Checklist for B2B Buyers

Use this checklist when sizing storage for a quote or comparing supplier bundles. Each item is a number you can demand in writing.

  • 1. Count cameras and their resolution/fps. List every camera, its resolution grade and frame rate — mixed sites need a per-channel calculation, not one average.
  • 2. Lock the codec to H.265+. Confirm the recorder decodes H.265+ per channel, not just headline support; it halves the drive bill versus H.264.
  • 3. Compute per-camera GB/day from the table. Use the H.265+ column for your resolution (6-10 GB/day at 4MP) and multiply by the scene activity factor.
  • 4. Multiply by cameras and retention days, then add 20-30% headroom. Days = usable capacity ÷ (cameras × GB/day); a 16-camera 30-day site needs roughly 4-6 TB.
  • 5. Choose bay count and RAID level together. 2 bays = RAID 1, 4 bays = RAID 5 (3/4 usable), 8 bays = RAID 5 or 6; never RAID 0 on footage.
  • 6. Buy surveillance-rated CMR drives. 24/7 duty, roughly 180 TB/year workload, short error recovery — and confirm the recorder's max drive size per bay first.
  • 7. Add a UPS. A power cut mid-write corrupts the file system and can drop a RAID array; a small UPS sized for the recorder is cheap insurance.
  • 8. Plan the backup and export path. Decide how footage leaves the recorder (USB export, eSATA, network), and verify a week of retention with a live test before handover.

For integrators buying at volume, add three supplier questions: the firmware update policy (recorders are computers), the RMA process for drives, and the compatibility list for third-party ONVIF cameras. A supplier who answers all three without hedging is one you can build a product line on.

Installer deploying a complete CCTV camera system

CCTV Storage Questions Buyers Ask

What does CCTV storage actually mean?

CCTV storage is the combination of recorder (NVR or DVR), hard drives and recording settings that decides how much video you keep and for how long. It is sized by per-camera GB/day × camera count × retention days, plus headroom — the math this guide walks through.

How long does CCTV footage stay on a recorder?

How long CCTV footage stays on your recorder depends on drive capacity divided by daily write volume. A 4 TB recorder with 4 cameras at 10 GB/day holds about 100 days; the same recorder with 16 cameras holds about 25 days. Once full, most recorders overwrite the oldest footage automatically.

How much storage does a 4MP camera need per day?

At 15 fps on a mixed scene, roughly 6-10 GB/day with H.265+, 9-13 GB/day with H.265, and 18-25 GB/day with H.264. The codec choice matters as much as the resolution — H.265+ cuts the H.264 figure by 60-70%.

Is 1 TB enough for a CCTV system?

For one 4MP H.265+ camera, 1 TB holds about 100 days. For 8 cameras it holds about 12 days — too short for most businesses. As a rule of thumb, a multi-camera site should start at 2-4 TB per recorder and scale with camera count and retention target.

Do I really need surveillance-rated hard drives?

Yes for any 24/7 system. Surveillance drives are rated for around 180 TB/year of continuous writes, use ATA streaming, and recover from read errors in seconds instead of stalling the recorder for minutes. Desktop drives fail faster under CCTV workloads and can drop out of RAID arrays.

Which RAID level should an NVR use?

RAID 1 on 2-bay recorders (50% usable, survives one drive failure), RAID 5 on 4-8 bay recorders ((N-1)/N usable), RAID 6 on 8-bay units with large drives. Avoid RAID 0 — it doubles capacity but halves your chance of keeping footage.

Can I store CCTV footage in the cloud instead of local drives?

Only for small or event-based workloads. Sixteen cameras at 2 Mbps need 32 Mbps of sustained upload, more than most business connections provide. The practical architecture is local continuous recording plus cloud events and clips for off-site backup.

The Bottom Line

CCTV storage is decided by three numbers (per-camera GB/day, camera count and retention days), and every other decision follows from them. Use H.265+ recording to halve the bill, size the hard drive size from the table (a 4MP camera: 6-10 GB/day), multiply by cameras and days with 20-30% headroom, protect the array with RAID 1 or 5 on surveillance-rated CMR drives, and keep continuous recording local with cloud as an event backup. A site that follows this plan keeps the footage it promises, survives drive failures, and never overwrites the moment a claim depends on.

At FULLCOLORCAM in Guangzhou we build 5MP F1.0 full-color night vision IP cameras, H.265+ NVRs and DVRs, and complete CCTV kits — every unit tested before shipping, ONVIF-conformant, with free P2P remote access and no subscription. Send us your camera list and retention target and we will return a spec-matched storage plan: recorder bays, drive size, RAID level and per-channel settings. Email cj226144@gmail.com for the calculation and a factory-direct quote.

NEXT STEP

Ready to Upgrade to Full-Color Night Vision?

Send us your requirements — our team responds by email.

RC

Written by

Ray Chan

CCTV Buyer's Guide Author · Full-Color Night Vision Specialist. Ray helps global importers and integrators source factory-direct security cameras.

← Back to Blog