I am currently comparing open-source tools for our startup's internal knowledge base. Flowise seems very flexible with its drag-and-drop node interface, but I am worried about the scalability when handling thousands of PDF documents. Has anyone successfully deployed a high-traffic app using Flowise in a production environment? I need to know if it handles concurrent user sessions reliably without memory leaks.
3 answers
I’ve been using Flowise for a production support bot for the last six months. The node-based approach is incredibly intuitive, especially for visualizing how the memory and the vector store interact. Regarding your concern about scalability, it really depends on how you host it. We use a high-memory Docker container on AWS, and it handles our 500 daily active users quite well. The key is to offload the heavy lifting to your vector database like Pinecone or Milvus. Flowise is just the orchestrator, so as long as your database is optimized, the UI shouldn't be the bottleneck for your RAG performance.
Are you planning to use the built-in SQLite database for persistence, or are you connecting to an external PostgreSQL instance for better stability?
The ability to export your flows as JSON and version control them in GitHub is a huge plus for our DevOps team.
Exactly, Susan. Managing Flowise deployments via CI/CD pipelines makes it feel much more like a professional software development tool rather than just a simple prototyping toy.
We definitely moved to PostgreSQL. The default SQLite in Flowise is fine for local testing, but for production-level concurrency, you really need a more robust database to manage your chat histories and session data without corruption.