L'écureuil volant : quand le CMS Spip devient serverless

In 1996, the first web hosting services emerged in France, such as Mygale and Chez.com. In 1999, the first content management systems (CMS) like Drupal and Spip appeared, allowing non-developers to become editors. By 2000, MySQL databases were introduced, but content publishing was still limited. By 2003, WordPress and Dotclear gained popularity, making it easier to write without developing.

By 2016, the author discovered cloud computing and was impressed by AWS's ability to provide highly available and scalable applications on demand. In May 2026, a call came in from Paul, who wanted to develop an application using Spip as the backend. Despite the requirement, the author and Paul aimed to create an application available to a wider public.

The challenge was to run Spip without incurring 24/7 hosting costs or maintaining servers during periods of low traffic. The solution was serverless architecture. After 6 days, Paul created SPIP-serverless, a serverless port of Spip. The project is available on GitHub under the MIT license and can be viewed on a demo site. SPIP-serverless functions by hosting the computation on AWS Lambda's Function-as-a-Service (FaaS), using Bref to run PHP on Lambda, sessions stored in DynamoDB, uploaded media on S3, CDN CloudFront for caching, and secrets stored in SSM Parameter Store.

The database is powered by Aurora DSQL, a serverless PostgreSQL, eliminating the fixed cost of always-on instances. While the serverless approach presents a compromise in terms of administration, the code is open-source and can be deployed as is. The author plans to maintain and improve the project based on real-world needs and welcomes contributions to develop and deploy custom SPIP projects.