



The application utilization for larger server farms in today's data centers is typically highly volatile due to many factors. Time of day usage patterns, promotions and advertisements, and unexpected events all cause spikes in demand, both planned and unplanned. As a result, data center operators have to deal with a high level of uncertainty when planning capacity. They mitigate the risk of a service outage by over-provisioning equipment and keeping their entire pool of servers "always on". The result is an average utilization below 20% with occasional peaks of 85-90%.
Case Study |
Presentation |