Databricks introduced generally available high-QPS scaling for AI Search, allowing customers to scale search endpoints to thousands of queries per second without manually managing replicas, infrastructure sizing or load balancers.

The capability targets production workloads such as real-time search, recommendations, personalization and entity resolution. Customers can specify their required throughput and Databricks automatically provisions the underlying infrastructure, while providing built-in monitoring for query volume, latency and endpoint health.

Source: databricks.com