Code Orange: Fail Small — Our resilience plan following recent incidents

Cloudflare has launched a resilience initiative named Code Orange: Fail Small, aimed at making the network more robust after two outages earlier in the year. On 18 November 2025, Cloudflare’s network failed to deliver traffic for about two hours and ten minutes, and on 5 December 2025 it again could not serve 28% of applications behind its network for around 25 minutes.

The plan organises work into three main areas: introducing controlled rollouts for configuration changes, reviewing and testing failure modes across systems handling traffic, and changing internal break-glass procedures to remove circular dependencies so teams can act quickly during incidents. It also emphasises aligning configuration updates with the controlled software release model, using mechanisms such as Health Mediated Deployments to define success metrics and automatic rollbacks if a step fails.

Cloudflare notes these updates will be iterative rather than a single big change, with goals to cover all production systems by the end of Q1 and to ensure safer, faster emergencies handling. The firm acknowledges the outages were painful for customers and the Internet, and promises ongoing updates as work progresses.