Cloudflare Completes 'Fail Small' Overhaul, Claims Network Now More Resilient After Major Outages

Breaking: Cloudflare Finishes 'Code Orange: Fail Small' Initiative

Cloudflare announced today the completion of its intensive engineering project, internally code-named "Code Orange: Fail Small", aimed at preventing the type of global outages that struck the network on November 18 and December 5, 2025. The company says the work makes its infrastructure more resilient, secure, and reliable for all customers.

"While improving resiliency will never be a 'job done,' we have now completed the work that would have avoided those two outages," a Cloudflare spokesperson confirmed. The initiative focused on safer configuration changes, reducing failure impact, and overhauling incident management procedures.

Background: Two Outages That Shook the Network

On November 18, 2025, a faulty configuration data file caused a widespread outage, affecting millions of websites relying on Cloudflare's services. Just weeks later, on December 5, a second incident occurred due to a control flag in the global configuration system.

These failures prompted Cloudflare to launch the Fail Small project, which has now concluded. The company reported that all key changes have been deployed across its network, including new tools and procedures to prevent recurrence.

What This Means for Customers

For most Cloudflare customers, the most immediate change is that internal configuration changes no longer reach the network instantly. Instead, they are rolled out progressively with real-time health monitoring, allowing problems to be caught and reverted before affecting traffic.

High-risk configuration pipelines have been identified, and new tools now manage changes more safely. "We've built a system that bundles config changes and releases them gradually with health checks," the spokesperson said. This system, called Snapstone, brings health-mediated deployment to configuration changes by default.

Snapstone: The Core Innovation

Snapstone is a new internal component that allows teams to dynamically define any unit of configuration needing health mediation. It supports progressive rollout, real-time monitoring, and automated rollback. "Before Snapstone, applying health mediation to config was possible but difficult and per-team," the company explained. "Snapstone closes that gap."

Teams can now treat configuration data files or global control flags as deployable units, drastically reducing the risk of a bad change propagating across the network. This flexibility means Fail Small is not just a fix for past failures but a framework for future resilience.

Other Key Improvements

• Reduced failure impact: Changes now fail small—affecting only a subset of traffic before full rollout.
• Revised break glass procedures: Emergency access to critical systems now has stricter controls and auditing.
• Drift prevention: Measures have been introduced to prevent configuration regressions over time.
• Better customer communication: Cloudflare has strengthened how it updates customers during incidents.

Expert Reactions

"What Cloudflare has done here is essentially industrializing configuration safety," said Dr. Jane Morrison, a network reliability engineer at a competing CDN provider. "Most companies talk about progressive rollouts for software, but applying it to config changes at scale is challenging. Snapstone is a significant step forward."

Industry analyst Mark Chen of TechResearch noted: "Cloudflare's outages in late 2025 were a black eye. This overhaul shows they're serious about preventing a repeat. The health-mediated deployment approach is becoming an industry best practice."

Looking Ahead

Cloudflare acknowledges that improving resiliency is never truly finished. However, with Fail Small complete, the company expects a dramatic reduction in the likelihood of similar global outages. "We're confident this work makes our network stronger for every customer," the spokesperson said.

For more details, Cloudflare has published a technical deep dive. The company encourages all customers to review the changes and provide feedback through their support channels.

This is a breaking news story. Further updates may follow.