IP Networks for CCTV

Analogue CCTV was an electrical job. IP CCTV is a networking job with cameras attached, and that shift is where a lot of otherwise capable installers come unstuck. The symptoms are familiar: cameras that drop out at busy times, recordings with gaps nobody can explain, a system that worked at commissioning and degraded once the client put other traffic on the same switch.

Bandwidth is the first constraint

Every camera produces a continuous stream. Total those streams at their peak bitrate, not their average, and compare against the capacity of every link the traffic crosses. The link that matters most is rarely the camera connection - it is the uplink between switches, and the connection into the recorder.

Sixteen cameras on a switch with a single gigabit uplink can saturate that uplink during busy periods even though each individual camera is nowhere near capacity. When it saturates, frames are dropped, and dropped frames become gaps in the recording that nobody notices until footage is needed.

Design against the peak. Scenes get busier, people add cameras, and bitrates rise when the picture is full of movement, which is exactly when the footage matters.

Power over Ethernet

PoE removes a separate power run per camera, which is most of why IP systems are quicker to install. The trap is the budget.

A switch has a total power budget shared across all ports, and it is usually well below the sum of what every port could deliver. A switch advertising PoE on all ports may only support a fraction of them drawing maximum power at once. Cameras with heaters, illuminators or motorised lenses draw far more than a basic fixed camera, and they draw it exactly when the weather is bad.

Cable length matters too. Voltage drop over a long run reduces what actually arrives at the camera. A camera that works on the bench and reboots intermittently on site is often a power problem masquerading as a network fault.

Keep the cameras separate

Putting cameras on the same flat network as office computers causes problems in both directions. Camera traffic is constant and heavy, and it will compete with the work the business actually does. In the other direction, cameras are frequently the weakest devices on any network - infrequently patched, sometimes running firmware the vendor stopped updating years ago.

A separate VLAN for the security system is the standard answer. It contains the traffic, and it means a compromised camera is not sitting on the same broadcast domain as the finance workstation. Where remote access is needed, it should go through a controlled route rather than by exposing the recorder to the internet, which remains one of the most common ways these systems are breached.

Addressing and documentation

Static addressing, or DHCP reservations, for every camera and recorder. Cameras that move address after a power cut turn a five-minute fault into an afternoon. Record what is on each port, in a document that lives somewhere other than the installer’s laptop.

Change the default credentials on every device, including the ones nobody logs into. Default passwords on recorders and cameras are actively scanned for.

Training

The CCTV and IP Networks bundle covers both sides of this - camera work and the networking underneath it - which is the combination most installers need. For the networking on its own, there is IP Network Fundamentals.

For the camera side, see CCTV camera selection and placement.