Unless you have a business relationship with whoever you rely upon then you are in trouble. Even then you should be ready to jump ship. Remember when S3 and EC2

Unless you have a business relationship with whoever you rely upon then you are in trouble. Even then you should be ready to jump ship. Remember when S3 and EC2 went down? If all of those services that rely on it were ready to jump onto another service they wouldn't have gone down as well. While it's not always possible to replace parts of your infrastructure you rely on, it's common sense to have multiple redundancies for stuff. The problem in the case of Twitter is their offering is users, rather than computing. That's something you can't replace if they pull the plug.