How to prevent an AWS lambda function from running more than once simultaneously?

AWS Lambda currently doesn't support this behavior. You could try using DynamoDB to create a lock, but that's obviously not the most ideal scenario. Right now this use case isn't a very good fit for Lambda.

I have a job that I need to ensure only one is ever running, and I wasn't happy with any Lambda solution for that. I ended up just running it on a t2.nano instance.

AWS Lambda FAQ says that the service is designed to run functions in parallel.

However, you could manage this from your application if you are using the microservice-http-endpoint blueprint. You can do it since this kind of blueprint is just a web service that you can call somewhere.