Microsecond Latency at Terabyte Scale: Inside Valkey Search

Valkey Search enables low-latency search and aggregation on large operational datasets for use cases such as real-time personalization and anomaly detection. In this talk, we dive into the architecture behind Valkey Search and examine how search workloads behave differently from traditional Valkey access patterns. We will walk through how queries execute across shards and within shards, and how the module’s multithreading and coordination model balances throughput, latency, and freshness. Attendees will leave with practical knowledge on cluster topology, query patterns, and the performance and consistency tradeoffs that matter when running search on hot data in production.