Ask HN: How do you monitor your infrastructure?

Hello! In the past day I've been researching how to do proper monitoring to our infrastructure, but unfortunately this seems much harder then expected. We need to monitor: 4x AWS Lambdas (errors & invocations), 2x Cloudfront statistics, CPU usage of EC2 instance, requests per second of heroku instance and its response time and few stats about Mongo instance.

If we were running only on AWS stack, Cloudwatch fit our needs, but we have some external dependencies.

I would like to hear, how do you monitor your infrastructure at scale and what tools are you using?