This paper presents a performance model of the ATA/ATAPI SSD Trim commandunder various types of user workloads, including a uniform random workload, aworkload with hot and cold data, and a workload with N temperatures of data. Wefirst examine the Trim-modified uniform random workload to predict utilization,then use this result to compute the resultant level of effectiveoverprovisioning. This allows modification of models previously suggested topredict write amplification of a non-Trim uniform random workload under greedygarbage collection. Finally, we expand the theory to cover a workloadconsisting of hot and cold data (and also N temperatures of data), providingformulas to predict write amplification in these scenarios.
展开▼