SD Times

ScyllaDB Releases Integrated Vector Search: 1B Vectors with 2ms P99s and 250K QPS Throughput

ScyllaDB announced the general availability of its integrated Vector Search feature within ScyllaDB X Cloud, targeting real‑time AI workloads. The service combines ScyllaDB’s shard‑per‑core architecture with a Rust‑based USearch extension to deliver approximate‑nearest‑neighbor indexing. In benchmark tests on the yandex‑deep_1b dataset, the platform achieved sub‑2 ms P99 latency and roughly 250,000 queries per second across a 1‑billion‑vector dataset. The design stores vectors alongside structured data, while a dedicated Vector Store builds in‑memory indexes, allowing independent scaling of storage and search layers.