Booking.com Reveals Why It Chose Weaviate Over OpenSearch as Its Vector DB

Booking.com's ML and DS teams detailed the process of switching their vector database from OpenSearch to Weaviate. This is not merely a tool change but a critical infrastructure decision supporting generative AI-based user experiences, including RAG (Retrieval-Augmented Generation) and semantic filtering.

The Evolving Role of Vector Search

In the past, vector search was treated as a backend implementation detail, but it has now emerged as a core element for improving personalization and accuracy across the platform. Booking.com emphasized that selecting a vector DB is an infrastructure decision of equal importance to choosing a primary data store or message queue, requiring predictable and scalable performance.

Migration Background

As embedding and vector search use cases surged across various departments within the team, Weaviate was adopted as the next-generation vector DB to overcome the limitations of the existing system and efficiently search more diverse domain data. Through this, the system aims to secure deeper context and improve not only search speed but also the quality of search results (accuracy and personalization).