Security Cameras & Video Surveillance

Every camera project should begin with two questions: how long do you need to keep footage, and what do you need to be able to prove when you go looking for it. Thirty days of continuous 4K across twenty cameras is a completely different storage build than fourteen days of motion-triggered 1080p across six. That answer sizes the recorder, the drives, and the network before anyone climbs a ladder.

Retention is often driven by something outside your control. Insurance carriers, franchise agreements, and liquor licenses all bring their own expectations. Claims also have a habit of surfacing weeks after the fact, long after a short recording loop has overwritten the only clip that mattered.

We size arrays with headroom instead of counting on compression to bail us out, spec surveillance-rated drives built for constant writes, and configure the recorder to actually alert someone when a disk drops out.

A camera that can detect movement across a parking lot is not the same camera that can identify the face at your back door. Detection, recognition, and identification each require different pixel density on target, and that math drives lens selection and mounting distance. Six cameras that answer real questions beat fourteen that produce wide, unusable overviews.

Scene conditions decide the rest. A glass storefront facing east needs serious wide dynamic range or you get silhouettes every morning. Tight interior corners with reflective surfaces punish infrared, so a turret usually outperforms a dome. Gates and loading docks generally want a dedicated plate camera at a shallow angle rather than one camera trying to do two jobs badly.

We walk the property before we quote it. Sightlines, existing lighting, sun position, ceiling construction, and where the tree line will be in July all change the design.

Camera systems fail at the infrastructure layer far more often than at the camera. New runs get Cat6, the 100-meter copper limit gets respected instead of treated as a suggestion, and fiber or a remote switch closet handles buildings that are longer than copper allows. Terminations get tested, not assumed.

PoE budget gets calculated on paper. A switch that advertises PoE on every port often lacks the total wattage to power every port at once, and PTZ units, integrated illuminators, and cold-weather housings draw far more than a fixed dome. An undersized budget shows up as cameras rebooting in January, which is a miserable fault to chase in the spring.

Cameras belong on their own VLAN with firewall rules governing what they can reach. That keeps camera traffic away from point-of-sale and business systems, and keeps an internet-connected device from becoming a path into everything else. Recorders go on a UPS, because a recorder that loses power mid-write is how a database gets corrupted.

We deploy UniFi Protect regularly and it earns its place. No per-camera licensing, a genuinely usable mobile app, and console options that scale from a single office to a multi-building campus. For a business that wants good video, simple sharing, and predictable cost, it is frequently the correct call.

It is not the only call. Some sites need ONVIF third-party cameras, specialized optics, analytics Protect does not carry, or a VMS that correlates video with access control events and register transactions in one timeline. Larger properties sometimes need failover recording or multiple recording servers.

Because we are not tied to one product line, the recommendation comes from your building and your requirements rather than from what happens to be on the shelf.

Remote access should never mean forwarding a port to a recorder. We use vendor-managed relay services or a VPN, keep firmware current, and set unique credentials on every device rather than leaving factory defaults in place. An exposed NVR is one of the fastest ways to end up scanned, brute-forced, and enrolled in somebody else's botnet.

Accounts get built around roles. A general manager might need live view and export rights, a shift supervisor might only need the sales floor, and ownership needs everything across every location. We also walk people through exporting a clip properly, because the moment you need footage for a police report is a bad time to learn the interface.

We stay available for support afterward, and we do not build dependence on us into the configuration. What labeling and documentation a job includes varies with its scope — that gets spelled out in the proposal rather than left to assumption.