An IP camera system is a set of network devices that capture video, compress it and send it to software that records and displays it. Designing one means choosing the right camera for each job, then sizing the network, power, storage and software so that every camera records reliably, all day and every day.
The parts of an IP camera system
Whatever the size of the site, the same building blocks appear:
- Cameras. Each one has an image sensor, a lens, a small computer that compresses the video and a network port. Most draw their power from the network cable.
- Cabling and PoE switches. Category cable runs from each camera to a switch in a telecom room. The switch carries the video and supplies Power over Ethernet.
- The core network. Uplinks, usually fibre between floors or buildings, bring video from the edge switches to the servers.
- Video management software (VMS). The software that records, indexes, displays and exports video, and manages users and permissions.
- Recording servers and storage. Either a purpose-built network video recorder (NVR) or servers with disk arrays. Some cameras also record to a memory card as a backup.
- Clients. Workstations, video walls and mobile apps used to watch live video and review recordings.
In a multi-storey building these parts are arranged much like any structured cabling system: cameras on each floor home-run to that floor's telecom room, and the floor switches uplink to the main equipment room where the recorders live.
Start with requirements, not cameras
The most useful document in a camera project is a short list of what each camera must show and why. For each location, write down the threat or question (who came through this door, which vehicle entered the yard, what happened at this till), whether the video will be watched live or reviewed later, and the level of detail needed. Detail is usually described as detecting, observing, recognizing or identifying, and each step up needs more pixels on the target. Our guide to camera placement and pixel density explains how to turn those words into a camera position and lens.
The same list should say who uses the system, how quickly incidents are normally discovered (which sets retention), which other systems need to trigger video, and who in IT will own the network and servers.
Choosing cameras
- Fixed dome and bullet cameras cover one view all the time. They are the backbone of most systems.
- Pan-tilt-zoom (PTZ) cameras can be steered and zoomed by an operator or follow a preset tour. A PTZ records only where it is pointed, so it complements fixed cameras rather than replacing them.
- Multi-sensor cameras put two to four sensors in one housing, covering a corner or a large open area from one mounting point and one cable.
- Fisheye cameras give a full overhead view of a room. They show movement well but little detail at the edges of the image.
- Thermal cameras see heat rather than light. They detect people in darkness and at long range, which suits perimeters, but they do not identify anyone.
- Licence plate cameras use fast shutters and infrared illumination tuned to read plates in a narrow lane.
Beyond the type, check low-light performance, wide dynamic range for scenes with bright windows or doorways, the operating temperature range and heater for outdoor use in a Canadian winter, and whether your chosen VMS supports the model fully.
The network and power behind the cameras
Each camera sends a continuous stream. The bitrate depends on resolution, frame rate, compression (H.264 or the more efficient H.265) and above all on the scene: a busy entrance with moving people, trees or rain produces far more data than a quiet corridor. Design from the camera maker's bitrate estimates for similar scenes, then confirm with real measurements after installation.
Main streams and sub-streams
Most cameras send two or more streams at once. The main stream carries full resolution to the recorder. A lighter sub-stream feeds live viewing grids, where a screen of sixteen tiles does not need full detail, and mobile apps on slower connections. Setting this up properly keeps workstations and remote links from being swamped.
Power over Ethernet
PoE power is defined in classes under IEEE 802.3af, 802.3at and 802.3bt, each delivering more power than the last. Fixed indoor cameras need little; PTZ cameras, outdoor cameras with heaters and cameras with built-in infrared illuminators need much more, especially on a cold night. Add up the worst-case draw for every camera on a switch and compare it with the switch's total PoE budget, not just the per-port rating. Ethernet channels are limited to 100 metres, so long runs to gates and poles may need fibre and a local power supply. Our guide to Power over Ethernet covers the classes and budgets in more detail.
Put the camera switches and recorders on a UPS. Without one, a brief outage restarts every camera, and the minutes they take to come back are minutes with no recording.
A worked example
Take a three-storey office with 20 cameras per floor and a few in the parkade, about 64 in all. Suppose the camera maker estimates around 4 Mbps for the main stream in these scenes. Each floor switch then carries about 80 Mbps of camera traffic, comfortably inside a 1 Gbps uplink. The recording server receives around 256 Mbps in total, which it must write to disk continuously while also serving playback. That figure, multiplied by the retention period, drives the storage calculation, and it is why a recorder that looks fine on a spec sheet can struggle when every camera is busy at once. A busy lobby or a rainy night can push bitrates well above the estimate, so leave headroom.
Securing the system
Cameras are network computers, and they are often the least maintained devices on a network. Basic measures go a long way:
- Put cameras and recorders on their own VLAN, separated from the office network by a firewall. Our guide to VLANs and network segmentation explains how.
- Change every default password and use individual user accounts in the VMS.
- Keep camera firmware and VMS software up to date, on a schedule.
- Do not open ports on the internet firewall for remote viewing. Use a VPN or the VMS maker's secure remote service.
- Synchronise all cameras and servers to one time source, so footage from different cameras lines up and stands up as evidence.
- Log who viewed and exported video, and review the log.
How ehZee can help
We design IP camera systems from the first question about why you need cameras through to verifying each camera's view after installation. Our video surveillance design service covers camera layout, network and PoE sizing, storage and integration with access control and alarms. Drawings for permit are sealed by a licensed professional engineer where required, and installation is done by licensed electrical contractors and qualified installers.
Common questions
Is a standalone NVR enough, or do we need servers?
For a small site with a few dozen cameras, a purpose-built NVR is often enough. Larger systems, multiple sites and systems that need redundancy usually move to servers running a VMS, which scale more easily.
Can cameras share the office network?
They can share the physical switches and cabling if the design allows for the bandwidth and PoE, but they should sit on a separate VLAN with firewall rules between cameras and office devices.
Is higher resolution always better?
No. More pixels mean more bandwidth and storage, and many cameras lose low-light performance as resolution rises. Choose the resolution that delivers the detail each view needs, then place the camera well.