Milvus DB — Milvus Introduction

Vector database intorudction

What is Milvus

Milvus is an open-source vector database that is designed to provide state-of-the-art similarity search and analytics capabilities. It is powered by Faiss, NMSLIB, and Annoy, which are libraries for efficient similarity search and clustering of dense vectors.

Established in 2019, Milvus was developed with a single ambition: to store, index, and manage enormous embedding vectors produced by deep neural networks and other machine learning (ML) models.

An embedding vector is a mathematical abstraction representing unstructured data, like emails, IoT sensor data, Instagram photos, protein structures, among others. In mathematical terms, an embedding vector can be described as an array composed of floating-point numbers or binaries. Contemporary embedding techniques are employed to transform this unstructured data into these informative embedding vectors.

Milvus is a database uniquely fashioned to handle queries over input vectors, with the capacity to index vectors on a trillion scale. Unlike traditional relational databases that primarily handle structured data following a pre-determined pattern, Milvus is fundamentally designed to manage embedding vectors derived from unstructured data. An example of Milvus data store…