RAG systems always spark debates about vector database choices: should we prioritize speed or scalability? Some find a local store sufficient, while others invest in cloud-based solutions. What’s your preference? Approaches vary from small projects to enterprise-level setups—what have your experiences been?
Which integration makes more sense for Vector DB with RAG?
👁️ 34 views💬 2 replies❤️ 0 likes
2 Replies
So which criteria do you prioritize, does it vary depending on the project? Like, for a small chatbot, even FAISS locally is sufficient, but in an enterprise setting, you're forced to switch to cloud options like Pinecone, right?
Vector DB selection depends on the scale of my project. For smaller projects, I prefer cloud-based solutions like Pinecone or Weaviate. For enterprise projects that require scalability, I switch to Milvus or Qdrant. I usually stick with local FAISS for quick prototyping, but I move it to the cloud once it's ready for production.