Scale a Search to 50M Users: Spring Boot Interview Question

Your search endpoint serves 50 million users, the database CPU is stuck above 85%, and latency is around 800ms. How do you get it under 100ms? In this Spring Boot interview question, we walk through how to scale a user search endpoint backed by PostgreSQL: Why ILIKE '%john%' forces a full table scan Speeding up substring search with a pg_trgm trigram GIN index Switching to a B-tree index when most searches are prefix-based (john%) Adding pagination with LIMIT to protect memory, network and DB CPU Using a Bloom filter to skip the database for usernames that don't exist (40% of searches in this scenario) TIMESTAMPS 00:00 The scenario: 50M users, 10K requests/sec 00:00 Why the current query is slow 00:00 Trigram index for substring search 00:00 Prefix search and B-tree indexes 00:00 Bloom filters for non-existent usernames 00:00 Wrap-up and interview tips READ THE FULL ARTICLE https://javabulletin.substack.com/p/s... More real-world Java and Spring Boot interview questions every week: https://javabulletin.substack.com If this helped, like the video and subscribe for more backend interview prep. #SpringBoot #Java #SystemDesign #PostgreSQL #InterviewQuestions #BackendDevelopment