Why Milvus 2.1 Is Revolutionizing Vector Databases for Unstructured Data

Milvus is a new open‑source vector database that can manage massive structured and unstructured data, helping developers accelerate next‑generation data structures.

The latest open‑source vector database system Milvus has just released version 2.1. On GitHub it already has 11.5K stars, 1.6K forks, and is licensed under the Apache open‑source license.

According to Zilliz, the company behind the project, Milvus is built as a similarity‑search extension for cross‑industry enterprises. It is designed for distributed architectures and can scale easily as data volume and workload increase.

Milvus supports DML operations, offers storage at the trillion‑byte scale, and provides near‑real‑time vector search.

Zilliz recently announced several new contributions in Milvus 2.1:

Further bridging gaps between data pools

Eliminating data silos

Improving performance and enhancing availability

“The data‑silo problem is now better integrated, allowing enterprises to unlock the full potential of their data,” says James Luan, project maintainer and engineering director at Zilliz. “When it comes to unstructured data, legacy database solutions often add features as after‑thoughts, whereas Milvus was designed from the ground up for unstructured data and now offers more built‑in capabilities to unlock powerful, integrated data processing.”

After the 2.1 update, Milvus users can enjoy performance improvements that reduce search latency for million‑scale datasets to 5 ms, while deployment and operational workflows are further simplified.

As an infrastructure for unstructured data, Milvus introduces revolutionary technology by handling vector embeddings rather than just strings.

Looking ahead, the Zilliz open‑source team will continue building an ecosystem around Milvus, with projects such as the open‑source vector data ETL framework Towhee and the visualization tool Feder already emerging to support non‑structured data interaction.

Author: 场长