Understanding of Dify Knowledbases

Hello together,

currently I am experimenting with difys community edition. I rolled out dify on AWS using a cdk stack I found on the internet(GitHub - aws-samples/dify-self-hosted-on-aws: Self-host Dify on AWS). I try to understand how dify works in order to find a solution for my company. Right now I am struggling with the following: How does dify create Knowledge Bases? I read an article of Dify using TiDB for it (Dify.AI Consolidates Massive Database Containers into One TiDB). So when I create a knowledge base in my example - does dify use the Postgres with TiDB to store the vectors? Is my understanding wrong? Would really appreciate help on that.

@marlon1989

Hi, if you install Dify using the CDK in that repository, the vector DB which is used is pgvector. Aurora supports pgvector. TiDB is not being used.