Video storage and retention

The storage a camera system needs comes from four numbers: how much data each camera produces, how many cameras there are, how many hours a day they record and how many days the video is kept. Get those right, add protection against disk failure, and recordings will still be there when someone asks for them.

What decides how much storage you need

Every camera produces a stream measured in megabits per second (Mbps). That bitrate depends on the camera's resolution and frame rate, the compression it uses, and most of all on the scene. A quiet storage room compresses very well; a busy entrance, a tree moving in the wind or a rainy night produces far more data from the same camera. Camera makers publish bitrate estimates for typical scenes, and design tools can model them, but the real figures only appear once the cameras are running.

To turn a bitrate into storage, there is a handy rule: one megabit per second recorded for a full day fills about 10.8 gigabytes. That comes from 86,400 seconds in a day, divided by eight bits per byte. Multiply by the number of cameras and the days of retention, then allow for the disk protection and headroom described below.

A worked example

A building has 40 cameras. The design assumes they average about 4 Mbps each when recording continuously. Together they produce 160 Mbps, and 160 × 10.8 gives about 1,730 GB, or roughly 1.7 TB, every day. Keeping 30 days needs about 52 TB of usable space. If the disks are arranged so that one or two can fail without losing data, some raw capacity goes to that protection, and a sensible design adds a margin for busy days and for cameras added later. The installed raw capacity therefore ends up noticeably larger than 52 TB.

The same arithmetic shows where the savings are. If half of those cameras watch quiet areas and record only on motion, their share might fall to a fraction of the continuous figure. Cutting retention from 30 days to 14 roughly halves the total. Each choice has a cost in what you can go back and see, which is why retention is decided with the people who will use the video, not only by budget.

Settings that change the answer

How long to keep video

There is no single right retention period. Start with how quickly incidents at your site are normally discovered and reported. A retail store usually knows about a theft the same day; a property manager may not hear about damage in a parkade for a couple of weeks; an insurance claim may arrive later still. The retention period should comfortably cover that gap.

Privacy law in Canada pulls the other way. Organizations are generally expected to keep personal information, including video of identifiable people, only as long as it is needed for the purpose it was collected for, and then to delete it. A written policy that sets the retention period, explains the reason and is applied consistently addresses both concerns. Some sectors and some insurance policies set their own expectations, so check those as well. Our guide to video surveillance and privacy in Canada covers the privacy side.

Exporting and preserving evidence

Because the window rolls over automatically, footage of an incident disappears unless someone exports it in time. Make exporting a defined step in incident response, with named people allowed to do it. A good export:

• Covers enough time before and after the event, from every camera that saw it.
• Uses the VMS's native format with a digital signature or watermark, plus a standard copy and a player if needed.
• Carries accurate time stamps, which depends on every camera and server being synchronised to one time source.
• Is logged: who exported it, when, why and who received it.
• Is stored securely and deleted when the incident and any related claim or proceeding are closed.

Where the video lives

Most systems record to on-site servers or network video recorders with arrays of disks. Cameras with memory cards can record locally as a backup, filling gaps if the network or recorder goes down. Cloud recording moves storage off site, which suits small, spread-out sites but needs enough upload bandwidth for every camera and a clear understanding of where the data is held. Larger systems often combine on-site recording with a cloud or off-site copy of selected cameras.

Protect the recorders the way you would any important server:

• RAID. Arrange disks so that one or two can fail without losing video, and replace failed disks promptly. Surveillance disks run continuously, so failures are expected over the life of the system.
• Health monitoring. Have the VMS alert someone when a disk fails, a camera stops recording or retention falls below the target.
• Power. Put recorders and camera switches on a UPS, as described in our guide to UPS systems for buildings.
• Failover. For critical systems, a standby recorder can take over if a main recorder fails.
• Access control. Limit who can view, export and delete video, and keep the audit log.

Check retention after installation

A storage estimate is only an estimate until the system has run through a full retention period. A few weeks after handover, and again after any change to cameras or settings, check:

1. The oldest recording on each recorder, compared with the target. Record the date and the gap, if any.
2. The measured average bitrate of each camera, day and night, compared with the design figure.
3. Free space and disk health on each array.
4. That an export from an older recording plays correctly on a separate computer.

Repeating these checks on a schedule, for example as part of routine maintenance, catches a failing disk or a camera that has started producing far more data before it costs you footage.

How ehZee can help

We size storage from measured or modelled bitrates, the retention period you set and the protection your system needs, and we design the recorders, network and power to match. Our video surveillance design service also covers export procedures and verifying after installation that retention actually meets the target.

Common questions