Optimal selection of garbage collection dirty threshold to balance power consumption and space release

Shingled Magnetic Recording (SMR) drives have been commercialized in recent years to provide high density and large capacity data storage. However, due to their append-only write property, the data update is usually handled through a log-structured manner, which causes old data to become invalid and dirty. Garbage collection (GC) methods are therefore introduced to clean the dirty data to release the disk space. However, the necessity of GC process performed remains an issue, as the process may increase the power consumption and downgrade the performance. Thus our emphasis in this paper is placed on the decision of the process under different workload conditions and system settings for overall power consumption reduction and disk space saving. An analytical model is built to study various scenarios, and a simulator is further developed to verify the correctness of the model. The selection of parameters, such as GC garbage ratio threshold for GC process in SMR drive, can be guided with our model and simulation tool.