Hello!
It might vary depending on the case, but if I’m looking to manage 5 billion 32-dimensional embeddings, is it okay to operate Weaviate using docker-compose in a multi-node setup for a production environment?
Thank you.
Hello!
It might vary depending on the case, but if I’m looking to manage 5 billion 32-dimensional embeddings, is it okay to operate Weaviate using docker-compose in a multi-node setup for a production environment?
Thank you.
Hi @Sungwon_Jung ! Welcome to our Community
That should work. 5 bilion is a lot of vectors, so you will need some beefy hardware
We also have a helm chart in case you want to use K8s:
I have answered a similar question here, but with isolated containers:
Let me know if this helps
Thanks!
Hi, @DudaNogueira
K8s cannot be used, so it seems that we need to operate with docker-compose using a multi-node setup.
I will test further whether docker-compose with multi-node can be used in a production environment.
Thank you.