Size a database connection pool against the limit your instance actually enforces. Shows steady-state demand, headroom, and the spike a rolling deploy creates when old and new pods overlap.
Work out the default max_connections for any managed database instance on AWS RDS, Google Cloud SQL or Azure Flexible Server, using each provider's published formula. Compare all three at the same memory.
Derive shared_buffers, effective_cache_size, work_mem and innodb_buffer_pool_size from your instance memory, and see the worst-case total when every connection sorts at once.
Set a minimum vCPU and memory and see the smallest matching instance on AWS RDS, Google Cloud SQL and Azure Flexible Server side by side, with the connection limit each one gives you.
Browse every managed database instance class with vCPU, memory, memory per core, max_connections and the shared_buffers or innodb_buffer_pool_size its memory implies. Filter by the shape you need.
Calculate the right application thread pool size from CPU cores, target utilisation and how long each request spends waiting on I/O. Uses Little's Law and the classic cores × (1 + wait/service) formula.