Crescent Bay Towers

The starting point

ExGadgeT came in on AMC for a CCTV system that was already installed: 29 cameras across six towers, the lift cameras among them. The network those cameras sat on had no internet connection, and nothing about it was managed.

That produced the failure the building actually experienced. Addresses were assigned by hand, the same address kept being assigned twice, and cameras dropped off with no obvious cause. Every occurrence meant a site visit — not to fix a camera, but to work out which addresses were still free and untangle the conflict by hand.

The request that exposed it

The client asked for 28 more cameras. The new coverage was specific: the lift maintenance room, the lobby, the fire-check floor, the refugee floor and the area inside it, and the skyline floor.

There were no usable addresses left in the existing arrangement. Adding 28 cameras to a network that could not already keep 29 straight would have multiplied the problem rather than extended the system, so the network had to be dealt with before the cameras went up.

The change

We separated the cameras from everything else and gave them a network of their own — a dedicated 10.10.0.x range carrying the surveillance system and nothing besides. The instability stopped immediately: with nothing else competing for addresses, there was nothing left to conflict with.

Then we planned where the junctions went, rather than letting them follow whichever cable arrived first. Each junction point aggregates a defined part of the site before linking back, and the split is logical as well as physical — so a fault now points at a junction, instead of starting a walk down every cable and every camera in the building.

The result

The camera count went from 29 to 57, and the recurring calls and site visits largely stopped along with it. The AMC is still running — that has not changed, and it is not meant to. What changed is what the AMC is spent on.

Minor issues can often be diagnosed remotely now, and some are resolved on site by the building's own staff — restarting a switch, checking a junction — because the layout can be explained to someone who is not a network engineer.

• A camera network of its own — surveillance separated onto a dedicated 10.10.0.x range, with no addresses shared with anything else.
• Junctions planned, not accumulated — each one covers a defined part of the site, so a fault is isolated to a junction rather than traced cable by cable.
• Far fewer site visits — the recurring conflict calls that defined the first phase of the AMC largely stopped.
• Maintainable by the building — simple faults can be handled on site without waiting for us to get in a car.