Some questions about Columnar Backend


I have some questions about Columnar Backend.

(1) Will setting up a columnar backend in the workflow configuration or preferences improve performance on all nodes?

(2) Where can I set up the columnar backend on the KNIME server?

I’d appreciate it if you could help me.


Hi @hkim,
The answer is unfortunately: “it depends”. For systems with limited main memory, the columnar backend pretty consistently outperforms the older backend. If you have main memory in abundance, though, the older backend can still be faster due to extensive caching. It definitely helps a lot with the Python nodes, though. The columnar backend has to be enabled on a per-workflow basis. When you enable it for a workflow in your AP and upload it to the server, the columnar backend should be used there as well.
Kind regards,


This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.