In any case i'll explain what is happening in the video:
A single eucalyptus instance is running and he gives some type of load to the instance and it reaches a point where he automatically deploys another instance to help maintain the load.
After a while, he increases the load even further when a ec2 instance is created and i also help in maintaining the load. When he decreases the load, the instances gets terminated to save cose etc.,
My question is, what does he use to give fake load to the instances running and any pointers on how to determine it?
Is it possible to do it in python?
You can use Siege to test server performance. You can emulate HTTP load on an instance from the command line, with options like concurrent users (-c) and the duration of the connection (-t):
siege -c20 -t30 http://www.yoururl.com/
If you just want to generate load on the server, use a tool like Stress to create CPU/IO/memory load.
Use Siege responsibly.