Technological advances in networking, mass storage devices, processor and information technology have resulted in a variety of data services in diverse applications such as e-commerce, health-care, scientific applications, etc. While the cost of purchasing technology is becoming cheaper, the same cannot be stated about the cost of managing an information infrastructure. In order to reduce this cost, one needs tools that empower system administrators to explain and reason about a storage subsystem’s past performance, e.g., response time. Ideally, an administrator would employ these tools to speculate on both physical organization of data and hardware changes. With a hypothetical change, one may use the previously observed response times to quantify the expected enhancements. In this study, we investigate linear regression, a M/D/1 queuing model and SEER as three alternative techniques to estimate response time. All techniques enable an administrator to speculate on changes to the placement of data and its expected impact on response time. A choice between these techniques is a tradeoff between accuracy and space/computational complexity to estimate response time. In our experimental studies, SEER provides a higher accuracy by using more storage space and computational cycles.