flash cut

A flash cut is the abrupt switchover of a complex electrical system from one configuration to another, with zero transition period. The old system stops; the new one takes over immediately. In power distribution networks, telecommunications infrastructure, and industrial control systems, this means cutting over at a defined moment in time rather than gradually migrating load or functionality.

The term originated in telecommunications, where switching networks historically required instant cutover to avoid call loss and service interruption. A flash cut of a telephone exchange meant taking the old switch offline and bringing the new one live at a precise moment, often at night or during a maintenance window to minimize customer impact. The word "flash" captures the suddenness: one instant the system is here, the next it is there.

Flash cuts are inherently risky. If the new system fails or is misconfigured, there is no graceful fallback. Unlike a phased migration where traffic gradually shifts to new hardware or software, a flash cut puts the entire load on the replacement immediately. This is why flash cuts require extensive validation: load testing, failover testing, configuration audits, and dry runs. Many organizations run parallel systems for days or weeks before attempting the cut, comparing outputs to catch logic errors before they go live.

In practice, flash cuts are scheduled during low-demand windows: nights, weekends, or planned maintenance shutdowns. A window of 2 to 4 hours is typical for critical infrastructure cutoffs. The engineering team remains on standby to roll back if the new system is unstable. In modern cloud and software deployments, flash cuts have become less common; blue-green deployments and rolling updates allow safer switchovers with automatic rollback. Legacy industrial systems and utility networks, however, often have no choice but to cut over at a defined moment.

The decision to execute a flash cut versus a phased migration depends on system architecture and downtime tolerance. If the two systems cannot run in parallel, or if switching is stateful and costly (as with certain industrial process controllers), a flash cut may be the only option. For systems that can tolerate brief outages but not extended instability, a flash cut is often preferable to a prolonged rollout with intermittent faults.