Justifying SSD Storage in Enterprise Cloud Environments

For enterprise applications that deal with large scale of data, storage IO is oftentimes the performance bottleneck. SSD (Solid State Drive) is increasingly being adopted by companies/applications to alleviate applications' IO bottleneck. However, not every application/product is justified to migrate to SSD from HDD (Hard Disk Drive), as such migration will incur more business cost due to SSD's higher price. On the other hand, for many business-critical products that are indeed IO-bound when the underlying storage media is HDD, they should migrate their storage to SSD to address the IO bottleneck in order to satisfy the SLA (Service Level Agreements). To justify the migration from HDD to SSD, naive approaches such as "try-and-see" involve substantial cost and time. In this work, we aim to develop a practical storage recommendation algorithm to automate the task of justifying the migration from HDD to SSD by considering both the application performance and business cost. We propose an algorithm and build a tool which takes the input of application performance data and disk IO characteristics incurred by the application, and outputs the result of whether migrating to SSD is justified or not.