ScyllaDB has announced the general availability of its new Vector Search capability, bringing massive-scale vector search to real-time AI workloads with industry-leading performance and efficiency. Designed to support the largest AI models and datasets, the integrated vector search can handle up to 1 billion vectors with sub-2 ms P99 latency and throughput reaching ~250,000 queries per second, addressing the limitations of standalone vector databases that many customers found costly and complex at scale. Built on ScyllaDB’s hardware-optimized shard-per-core architecture with a Rust-based extension that leverages the USearch approximate-nearest-neighbor (ANN) library, this solution stores structured attributes and vector embeddings in a unified distributed table while a dedicated Vector Store service handles indexing via change data capture (CDC), allowing independent scaling of storage and indexing for optimized workload handling. This architecture achieves remarkable speed and predictable performance, making it suitable for real-time AI use cases such as latency-sensitive machine learning, predictive analytics, fraud detection, and other high-throughput similarity search applications.
Also Read: AWS Enhances Amazon Bedrock Data Automation with Blueprint Instruction Optimization
“ScyllaDB supports the most scalable vector search deployments at monstrous speed,” said Dor Laor, co-founder and CEO at ScyllaDB. “Based on publicly available benchmarks, ScyllaDB currently demonstrates the fastest vector search performance at billion vector scale. It also offers excellent TCO across all model sizes. That means teams can support and scale their largest AI inference workloads without the traditional performance-cost tradeoffs.” This vector search function, ScyllaDB Cloud, is intended to simplify architecture and eliminate the requirement for dedicated vector databases and with it lower overall cost of ownership, making it easier for teams to handle and scale demanding AI inference applications with little trade-off between performance and cost. This new ScyllaDB milestone has further reinforced the company’s commitment to delivering consistent scalability and performance capabilities with its real-time applications and expanding ScyllaDB Cloud capabilities.
