Hey
@Md.Riad Hasan Sarker ,
I took a look at your cluster's metrics, and you were hovering around the 2GB RAM mark a couple of hours ago. Typesense typically needs 2x-3x your dataset size in order to function properly. 2GB and especially 4GB should have been enough to serve your 800MB dataset. However, there's extra RAM usage when adding Vector Search into the mix. Are you using any embedding models for semantic search?
Another thing that I came across is that you're exclusively using the single writes endpoint for indexing documents, more than 50 times per second in some cases, which is going to lead to CPU usage spiking. In cases like these, using a buffer table and using the batch import API to flush that table in a 1-5 second interval will massively help with CPU usage. More info on indexing here:
https://typesense.org/docs/30.1/api/documents.html#index-documents