CCTV Storage: How to Work Out Days of Retention

Two questions decide every recorder purchase: how many days of footage you want to keep, and what each camera costs you per day in gigabytes. Drive count, recorder model and whether you need a second unit all come out of those numbers.

Bitrate is the only number that matters

Resolution, frame rate and compression do not consume storage directly. They consume storage through bitrate, which is how many megabits per second the camera sends to the recorder. Once you know the bitrate, the arithmetic is fixed.

One megabit per second, recorded continuously, uses about 0.45 GB per hour, which is about 10.8 GB per day.

So a camera streaming 4 Mbps around the clock writes roughly 43 GB a day. Ten of those cameras write about 430 GB a day, and thirty days of that is about 13 TB before you allow for anything else.

Estimates go wrong because most cameras run variable bitrate. A static dock at 3am produces far less data than the same camera watching rain, trees and forklifts at 11am, and manufacturer figures are usually the quiet-scene average. Build in headroom.

Typical starting figures

These are common ranges for business cameras at 12 to 15 frames per second with H.265. Treat them as a starting point and confirm against the actual camera.

Busy outdoor scenes sit at the top of each range or above it. A camera watching a car park full of moving vehicles in wind can exceed its quoted bitrate for hours at a time.

Frame rate: where the easy savings are

Most business footage does not need 25 or 30 frames per second. A common choice is 12 to 15 fps, which halves the data against 30 fps while still showing a person walking, a hand reaching, a vehicle passing.

Where higher frame rates earn their place:

• Till and point-of-sale cameras, where hand movements and note handling matter.
• Gates and driveways where vehicles are moving quickly.
• Plate capture, which has its own requirements — see the plate camera guide.

Below about 8 fps, people start jumping between positions, which is hard to rely on when establishing a sequence of events.

H.264 versus H.265

H.265 (also sold as HEVC, or in vendor flavours like H.265+ and Smart Codec) compresses the same scene considerably harder than H.264 — in practice often around half the bitrate for comparable quality, though the saving depends on the scene.

The trade-offs worth knowing:

• Older recorders and some third-party software do not decode H.265 well, or at all.
• Browser playback can be patchy depending on the device.
• Vendor "plus" variants achieve their savings partly by heavily compressing static backgrounds, which can smear detail when something finally moves across them.

For a new system, H.265 is the sensible default. If you are mixing old and new equipment, check what the recorder and the people who will review footage can actually play.

Continuous or event recording

Continuous recording writes everything. It is the safest option: the footage you did not know you needed is there, and you can scrub back without wondering whether a trigger fired.

Event recording writes only when something triggers, usually motion, a line crossing or an external input. On a quiet site it can cut storage dramatically — often to a fraction of continuous — which buys you more days of retention on the same drives.

The risk is the trigger. If the trigger is classic pixel motion, you inherit all of its weaknesses, and you will eventually find a gap where the threshold was set too high. Our guide to AI detection versus motion detection explains why.

A common middle ground: record continuously at a modest frame rate, and keep a separate higher-quality or event-flagged layer for the cameras that matter most. Many recorders also support a pre-record buffer, so an event includes the seconds before the trigger. Our own analysis layer captures short clips around each event, roughly ten seconds either side, which makes review quick without replacing the continuous archive.

Working out your retention

1. List every camera with its resolution, frame rate and codec.
2. Estimate the bitrate for each from the table above or the camera's own figures.
3. Multiply total bitrate in Mbps by 10.8 to get GB per day.
4. Multiply by the days you want to keep.
5. Add 20 to 30 per cent headroom for busy scenes and variable bitrate.
6. Compare that figure to usable drive capacity, not advertised capacity — a 10 TB drive gives you roughly 9.1 TB usable once formatted, and RAID reduces it further.

The storage planner does this arithmetic for you, and the camera count tool will give you the camera list to feed it.

How many days you should keep is partly a business decision — how long before you would notice a problem, how long an insurer or your own investigations need — and partly a matter of the rules that apply to your records. That is general information, not legal advice, and the laws overview explains why you should check your own state's position.

Choosing the recorder

Capacity is only part of it. Check:

• Channel count with room to grow. Buying a 16-channel recorder for eight cameras costs little and saves a replacement later.
• Incoming bandwidth limit. Every NVR has a maximum total Mbps it will accept across all channels. Eight 4K cameras can exceed a modest recorder's limit even though the channels are free.
• Drive bays and supported drive size. Four bays taking 10 TB drives is a different machine to two bays taking 4 TB.
• Surveillance-rated drives. Drives built for continuous write workloads, not desktop drives.
• RAID or redundancy, if losing the archive to one drive failure is unacceptable. RAID costs capacity and is not a backup.
• Export that produces a file somebody else can play, with a timestamp, without installing a proprietary viewer.

Recorders are also the part of the system most often left exposed to the internet, which is a serious and avoidable risk — our network guide explains what to do instead.

If you would rather hand over the sizing, ask for a quote and we will size the recorder, drives and switch against your camera list.