Unveiling Milvus 2.3: A Milestone Release Offering Support for GPU, Arm64, CDC, and Many Other Highly Anticipated Features

What this post added

Milvus 2.3 introduces GPU indexing using the RAFT algorithm via Knowhere, leading to 3-10x faster QPS. It adds Arm64 support for Docker images. The upsert operation is implemented as a delete-then-insert combination. Range search allows specifying distance bounds for queries. ScaNN index is now supported for faster queries. The Growing Index automatically indexes streaming data. An iterator interface in Pymilvus enables batch retrieval of over 16,384 entities. MMap support increases single-machine data capacity. CDC is introduced for data synchronization and backup. Performance improvements include optimized HNSW filtering, full multi-core CPU utilization, a refactored stateless QueryNode, and a new message queue system using NATS and Bookkeeper. The load balancer is optimized for faster fault detection and adjustment.