Rate limiting in AWS Lambdas

What if the lambda calling source is AWS API gateway and this gets called by the consumer, can we stop the lambda invocations in this case? Or if we can store the limit of allowed request in ddb access it before each of the call in lambdas and if the limit has got exhausted, we remove the connection between AWS API Gateway and lambdas

Is this possible?

Also in the case of API gateway talking with lambda and the command is given to make the reserved concurrency to 0, when does the concurrency becomes 0, will it becomes after all the requests from API gateway are served or how does it works?

But this solution will not work, can we stop the lambda from being invoked after a certain limit gets reached?

I can give you couple of ideas but none of them comes out of the box, you'll have to put a logic:

Option1: You can get aws/lambda -> Invocations for given duration through get-metric-data and as soon as that limits gets hit, no processing would happen however lambda would be invoked. This is something you would do within lambda function.

Option-2: Before this lambda function execution, another lambda function monitors that lambda function invocation and if let's say 50 invocations have already happened in last 30 minutes, then this lambda function would put a switch tag on that lambda function such as "Run":"Off" and your process flow would check this tag value first before invoking lambda function, if invocation count is less than 50 in last 30 minutes, toggle this switch back to on and invoke that function, otherwise don't invoke it.

You can leverage Cloudwatch alarm feature as well to achieve this use case.

I'm not sure if either of these options would be feasible for you based on complexity it'd add. I'd also encourage you to explore AWS State Machine(step functions).

Hope this helps.