The Evolution of Milvus - A Cloud-Native Vector Database

Explore the evolution of Milvus, a cloud-native vector database, in this informative conference talk. Delve into the challenges posed by the rapid increase in unstructured data and learn how embeddings can effectively represent and analyze this information. Discover the design principles behind Milvus 2.0, a fully distributed and horizontally scalable solution trusted by over 1000 organizations. Gain insights into the paradigm shift in vector database architecture, from version 1.0 to 2.0, and examine real-world applications. Understand key concepts such as unstructured data, embeddings, and vector databases, while exploring Milvus' features, cloud-native architecture, and general structure including access, coordinator, worker, and storage layers. Conclude with practical takeaways and diverse use cases, from art gallery experiences to threat detection, property search, and product recommendations.

Intro
What is Unstructured Data?
The Evolution of Data
Embed the World
Embeddings are Powerful
What is a Vector Database?
Vector Databases in Production
Milvus Features
Cloud Native
Vector Data ETL
Administrative and Visualization Tools
Milvus General Architecture
Access Layer
Coordinator Layer
Worker Layer
Storage Layer
Key Takeaways
Art Gallery Experience
Threat Detection
Property Search
Product Recommendation