The whole Web fraternity has been gaga over Web performance. Web performance can cure baldness. Jon Jenkins argues that even operations can. In other words, optimizations in Web operations can lead to significant cost savings and higher business profits.

The usual mindset of an operations manager is to plan capacity expecting, say, a 15% overload above peak. This approach works but focuses on spending. A new approach that focuses on saving is revealed.

There are two problems with the traditional capacity planning models. One, that you cannot do anything about hardware under-utilization during a slump. Every November till the end of December, Amazon had a 75% decrease in the demand for deployment servers. However, you couldn't throw away those servers during the time you were not using them. This resulted in a wastage of a couple of tens of millions of dollars.

Second, that even if you plan to purchase a fleet of servers that will meet your peak demand plus 15%, you cannot avoid times when your hardware need crosses that limit, too.

Besides, the step-up costs for increasing hardware are usually equivalent to the cost of an additional rack of servers, and not the cost of a single server. And once you have purchased the last rack, the marginal cost of an additional server is equal to the cost of a new data center, approximately $10 million dollars by common standards.

Jon Jenkins, the Director of Platform Analysis at Amazon argues that all these issues can be mitigated by using the Amazon utility computing platform, EC2.