How I Built a Serverless Search for My Blog

A developer built a serverless search feature for a Hugo-based static blog using Apache Lucene compiled to a native binary via Quarkus and GraalVM, deployed on AWS Lambda with API Gateway. Key challenges included making Lucene work in a GraalVM native image (requiring custom substitutions), achieving cold start times of ~180ms on Lambda, and managing AWS IAM permissions and cost control. A warm-up ping trick is used to pre-warm the Lambda when users focus the search input, eliminating perceived cold starts. The post covers data extraction via Hugo custom output formats, the JAX-RS HTTP service, deployment with AWS SAM, IAM policy configuration, performance tuning, and cost control using API usage plans with a budget-triggered Lambda to disable the API if spending limits are hit.