Skip to main content
  • EN

I have been working on some ideas involving RAG and Vector Databases, this initial post just explains what they are and how they can be used. 

 

https://vzilla.co.uk/vzilla-blog/my-thoughts-on-retrieval-augmented-generation-rag-and-the-power-of-vector-databases

The next steps is potentially a post on how do we protect these vector databases when bad things happen, the spoiler alert being these are like no other databases, they need protecting but also those backups can be used to do some interesting things. 

Try with Mongodb :) . That is what I did but to be honest for my setup something leaner and meaner would have worked better like pinecone. 

great post. You now can leverage a Kasten blueprint to backup the database :) As long as your postgres in in the Kube :) 

So an interesting next step is to cover the protection of that data… This is actually running on a system with a GPU but it could be in Kubernetes, this is a VM on vSphere which actually provides me the ability to not only protect this but also provide Instant DB recovery and the Explorers… which i do not get in Kubernetes. 


I did look at MongoDB for vector databases and I do not think it works as well as Postgres, then as you mention there are many PaaS offerings you can also leverage. I am reluctant to use these as you are sharing data with them. 

Terms & ConditionsAccessibility statement