You should implement this tutorial:
[login to view URL]
I got down to the section on node.js.
My requirements, beyond doing what it says in the tutorial are:
1. The stream be unfiltered, in other words, the entire twitter stream be downloaded.
2. The system can run unmonitored and not crash
3. Use [login to view URL] EC2 for the node.js code
4. I want this to be done at minimum cost and this I believe means moving the data from S3 to Glacier storage frequently and automatically.