
6/21/2022 · Dipanjan Sarkar
What this post added
This post introduces the fundamental concepts of vector databases and similarity search, explaining how unstructured data is represented as vectors (embeddings) and the challenges of storing, representing, and querying this data. It details the necessity of vector databases for handling massive amounts of vectors efficiently, outlining key requirements such as scalability, reliability, and speed. The post then introduces Milvus as an open-source vector database, highlighting its role in accelerating MLOps and its capabilities in storing, indexing, and querying embedding vectors, including support for attribute filtering using a bitmask mechanism.