Bandwidth is the network capacity available to move video and data — how many megabits per second the links between cameras, recorders, servers, and the cloud can carry. It is one of the first hard constraints in any surveillance design, because video is heavy: a single IP camera commonly produces 2–8 Mbps, so a few dozen cameras can saturate a link, and the upload side of an internet connection is usually far smaller than the download side.
Bandwidth shapes the architecture more than almost any other number. Local recording keeps the heavy streams on a fast LAN; sending continuous video to the cloud runs straight into the "upload wall", since 40 cameras at 2 Mbps need a sustained ~80 Mbps upstream that most business links cannot give 24/7. This is exactly why edge analytics and hybrid designs exist: sending compact metadata instead of full video can cut the required bandwidth by around 99%.
The pitfall is planning for capacity but forgetting sustained load and direction. Surveillance traffic is continuous, not bursty, and largely upstream — the opposite of typical office use — so a link sized on peak download figures can still collapse under round-the-clock camera upload. Size the network for the worst-case simultaneous streams in the direction they actually flow, and use substreams, edge filtering, and local recording to keep heavy video off the constrained links.
