Vector search integrates into existing database engines

The landscape of vector search is shifting as dedicated vector databases increasingly integrate into existing database engines. Major platforms such as Postgres, Elasticsearch, OpenSearch, ClickHouse, MongoDB, and Redis have implemented vector search as a data type, allowing users to perform approximate nearest-neighbor (ANN) searches alongside traditional queries without the need for separate, specialized systems.

While this integration reduces operational complexity, the economic challenges of scaling vector data remain. Costs are primarily driven by the memory footprint of dense embeddings and the efficiency of compression, regardless of the specific vendor.

MariaDB has also implemented vector search capabilities to support AI applications. However, Kaj Arnö, executive chairman of the MariaDB Foundation, noted challenges in gaining visibility within the AI ecosystem. He expressed concern that AI coding assistants and LLM-driven development workflows often default to other databases, potentially overlooking MariaDB due to training data limitations or existing documentation trends.