I’m new to GenAI and currently researching on which vectorstores are good. I came across FAISS and it’s good but it stores data locally but I’ve seen some vectorstores that stores the vectors on databases like AstraDB uses Cassandra. So if some vectorstores provides Databases why should anyone go for FAISS? Also if someone want to use FAISS, which databases should we consider?
FAISS is the fastest and most efficient vector similarity search tool, making it perfect for apps that need to run quickly. It’s open source and flexible, which are both good things.
Database-backed vector stores like AstraDB with Cassandra, make it easy to scale and come with extra features and built-in persistence. They may, however, perform a little worse and cost a little less.
For pure speed and freedom, FAISS is the way to go. Database-backed options, on the other hand, are better for scalability, data management, and ease of use.