The concept of a serverless application is designed to work very well with services, so establishing the connection to RDMBS within the function could be troublesome, as the parts of the services are still relied upon to provide the fast responses.
As function scales with RDBMS connections, the number of connections grows with the scale, and we might end up introducing a bottleneck and an I/O wait into the cold start of the function.
So Serverless architecture makes us rethink the data layer, so if we try using serverless with the existing data layer (RDBMS), then you will see a performance lag in your overall application performance. This might also be the reason why DynamoDB works so well with Serverless. ...
