Hi community,
I’m Darshan Hiranandani, Decheng, and I’ve successfully deployed Verba with a local Weaviate database on my machine. Since Weaviate operates in-memory, I’m looking to allocate more memory to the Weaviate process using NUMA (Non-Uniform Memory Access) for better performance.
I came across an example using llamaindex, where the vector store is managed in memory. In the example, when the vector_store is set to None, it creates an in-memory vector store that allows fast access but doesn’t persist data after the program ends. This method works by allocating memory directly to the vector store.
I’m wondering if a similar approach can be applied to Weaviate for memory management. Does this idea make sense, and can I use NUMA to assign more memory to the Weaviate process in a similar way?
Thanks for your help!
Regards
Darshan Hiranandani