How to start horizontal scaling in Typesense? As the data size grows it seems that the response time and ram are correlated to each other. Is there a possibility to split and data into two server and make the connection between them to give unified result?

Typesense supports a multi-node setup in a clustered mode, but each node in the cluster stores a full copy of the dataset. So you can scale read throughput horizontally by adding more nodes (at a slight expense of increase in write latency as you add more nodes). But if you want to scale the amount of data, you can only scale vertically by adding more RAM to each node in the cluster.

so the best way is to putt different collection in different cluster. if the data grows more.

I would recommend first increasing RAM on the same cluster to as high as your cloud provider allows. For eg, AWS offers up to 24TB of RAM on a single node. If your data exceeds that, then you’d provision separate clusters and then split data between them

My concern is about HA. So i am trying to setup multi node structure. @Jason Bosco

Ah I see, Typesense already supports a multi-node clustered setup: https://typesense.org/docs/guide/high-availability.html