What kind of performance can be expected at each tier?

Currently, I have an Atlas serverless cluster with 1 million documents, roughly 2GB.

If a query takes 10 seconds to run with my current environment, what kind of performance can I expect with each tier here?:


This question isn’t directly answerable as it depends on your overall concurrent workload and resources. If you are expecting changes in performance (for better or worse), I would start by trying to understand how your query is limited by current resources.

For example:

  • Have you profiled time spent in different areas of your application’s request lifecyle (eg database processing, network transfer, application processing, and frontend rendering)?

  • Are there significant memory, storage, or network limitations affecting your database throughput?

Typically performance improvements are going to involve tuning across all areas involved in the request lifecycle.

If the most limiting factor happens to be available RAM for query processing, upgrading to a larger database cluster tier would likely help. If the main performance challenge is time spent converting results from BSON (returned by the server) to native objects in your programming language or rendering in the frontend, changing your cluster resources is unlikely to improve that significantly.

There may also be some performance wins that would avoid scaling up database resources such as checking:

Hope that helps!