A Novel and Generic Model for Synthesizing Disk I/O Traffic Based on The Alpha-stable Process

Obtaining representative and concise I/O workloads for the purpose of projecting the performance of storage systems remains a challenge due to the complex nature of I/O behaviors. Previous studies have shown that disk I/O traffic can be represented as an independent and identically distributed random process in some workloads and a self-similar process in others. Additionally, workloads in the presence of self-similarity can exhibit either Gaussian or non-Gaussian characteristics. This paper proposes a new and generic model based on the alpha-stable process to accurately build a synthetic workload representative of I/O traffic in production storage systems. The novelty of this new model is that it has the capability of characterizing both self-similar Gaussian and non-Gaussian workloads. Experimental results show that this model can accurately capture the complex I/O behaviors of real storage systems and more faithful than conventional models, particularly the burstiness and heavy-tail characteristics under the Gaussian and non-Gaussian workloads.