An optimization model for shingled disk via hybrid mapping and multiple garbage collection strategies

The emergence of shingle recording technology has made the high-capacity disk possible, which does not change the original magnetic medium, but add a shielding layer to the original disk. However, random write performance of the current shingle disk is very poor, which becomes an important factor restricting the extention of its applications. We propose an Optimization Shingle Disk Model short for OSDM to provide a feasible method for shingle disk. Through the study of traditional shingle disk model, we list the following optimization aspects: 1. Redesign mapping module, aiming at the random write optimization, where we proposed the partition mapping using SR-Sector and SR-Block structures; 2. In garbage collection module, through the dynamic optimal and greedy algorithms, each region has shown good collection efficiency; 3. For data transfer module, a multiple collection has shown to improve the separation efficiency of cold and hot data. A prototype of OSDM also has been implemented by software simulation, and the results also show excellent performance as well as 5 times significant capacity.