Introducing Milvus 2.6: Affordable Vector Search at Billion Scale

What this post added

Milvus 2.6 introduces RaBitQ 1-bit quantization for significant memory reduction (up to 72%) and performance gains (4x QPS) without compromising recall. It also enhances full-text search with a revamped multi-language analyzer, adds Phrase Match for nuanced word order search, and introduces Time-Aware Decay Functions for relevance ranking based on age. A new Function interface integrates embedding models directly into the data pipeline for a streamlined 'Data-In, Data-Out' experience. The architectural evolution includes replacing Kafka/Pulsar with a new Woodpecker WAL system for improved performance and reduced operational complexity, and introduces tiered storage for cost reduction.