Date Added: Mar 2012
This paper proposed a cost estimation model for web applications on Amazon EC2 with auto scaling. The authors' approach was largely based on simulation with about 10 parameters. They observed that the cost was subjected to the distribution of users over time. The uniform distribution was the most costly. They found that cool-down time was the most promising parameter for cost reduction. However, cutting the cost lowered the quality of service. Finding an acceptable tradeoff was a key issue in parameter optimization.