Multiple backups. One shared point of failure.

01The challenge

Electricity disruptions rarely stay with one device or one application. For a wholesale distributor, an outage reaches whatever shares the same supply.

The business had backup arrangements in place. What it lacked was a consolidated picture of which components it operates and how a disruption would travel between them.

02The hidden dependency

The team used Failcast to record infrastructure components and the dependencies between them, the same structure Failcast uses to model a business service on the systems underneath it.

In that model, the backup power paths are not independent all the way down. Before reaching the equipment they protect, they converge on one component: a shared distribution panel. If it fails, power from the backup paths behind it may not reach what depends on it, however many of them are available. The redundancy addresses a failure of supply, not a failure of the distribution point.

Redundancy does not eliminate a single point of failure when independent-looking systems share a critical dependency.

Each path can look complete on its own. The shared dependency only shows when the paths are drawn together, with the dependent operations behind them.

03What the business gained

1. 01Dependency visibility
   A model of a substantial part of its infrastructure, with the relationships between components in one place.
2. 02Shared-risk identification
   One documented potential single point of failure in its backup power arrangement.
3. 03A reusable model
   A basis for examining further failure scenarios.

Beyond power infrastructure

The same pattern shows up far from electrical panels: two internet links through one duct, two cloud regions behind one identity provider, a recovery environment on the same DNS as production. A diagram that shows two of everything will not show the dependency they share. An inventory lists what you have. A dependency model shows what fails together.

Scope and methodThis case study covers one finding from one deployment. It does not report financial impact, downtime, or changes made to the installation, and it is not an independent assessment. The diagram is conceptual: the number of backup paths and the downstream systems are simplified.