How MarsBased website survived Cloudflare’s global outage

On November 18, 2025, Cloudflare suffered a major global outage that made headlines across the industry. For many businesses, the impact was immediate: websites down, assets unreachable, and traffic tanking. At MarsBased, Cloudflare had been a key part of our stack for years, and we still use it for several client projects.

It offers a compelling mix of performance, security, and convenience. We highlight its DDoS protection, fast edge routing, an excellent DNS service and powerful add-ons like image optimization and Workers. For teams building and scaling online products, it’s an incredible value.

Our past setup

Our website is currently hosted on Render.com. It delivers a smooth deployment experience, a friendly UI, and solid monitoring out of the box. On top of Render, we used to have Cloudflare’s proxy to enhance performance and handle image operations automatically.

This setup gave us:

• Automatic WebP conversion to serve lighter, faster images.
• Responsive image variants generated on-the-fly for different screen sizes.
• Global CDN acceleration for JavaScript, CSS, and static assets.
• Significant hosting cost reduction thanks to Cloudflare’s caching.

The unexpected problem: A legal dispute in Spain

Our issues began in February 2025 with an unrelated legal conflict between Cloudflare and Spain’s professional football league. Because Cloudflare is often used as a proxy (sometimes by sites distributing illicit football streams) the courts authorized the league to block certain Cloudflare IP ranges.

The side effects were dramatic. On match days for both La Liga and the Champions League, the MarsBased website became completely inaccessible from Spain. Since our traffic relies in part on a Spanish-based audience, this was a serious problem.

Eventually, the cost of lost traffic outweighed the convenience and performance benefits. We had to make a call: stay with Cloudflare and accept the risk, or restructure our stack.

We chose reliability.

Rebuilding without Cloudflare

Fortunately, Render had introduced a new feature earlier this year: Edge Caching for Web Services, allowing us to serve static resources from their own edge network. This covered part of Cloudflare’s role without adding complexity.

The remaining challenge was image processing. We evaluated several alternatives, Cloudinary and BunnyCDN among them, but ultimately landed on ImageKit.io. It offered the right mix of features, simplicity, and pricing, and the migration was refreshingly straightforward.

The result: we replaced Cloudflare’s image transformation and caching layer without affecting the user experience. And because of this change, when Cloudflare went down globally on November 18, MarsBased stayed online.

The bigger lesson: Every stack has dependencies

Incidents like this are a reminder that no matter how "cloud-native" or "distributed" your application is, you’re always depending on third-party providers. Cloudflare is a resilient, industry-leading infrastructure provider but no platform is immune. AWS is another recent example that had a global outage a few months ago.

As builders, we need to design systems that acknowledge this reality. Providers fail. Networks fail. Legal and geopolitical issues can have unexpected ripple effects. And sometimes the only way to reduce risk is to diversify your dependencies.

For us, decoupling key parts of our website from Cloudflare gave us more control and reduced the blast radius of external failures. It paid off sooner than expected.

If you’re considering how to make your own stack more resilient, or want to explore what migrating away from a single point of failure could look like, we’re always happy to help.