Contention Root Cause Analysis Framework of Cloud Distributed Storage

Competition for resources in storage systems can lead to a decline in cloud Service-level Agreements(SLAs). This paper proposes a cloud storage resource contention analysis framework, which can identify storage performance problems efficiently and accurately, and locate the load of the largest volume causing the storage performance problems. Firstly, the storage performance detection model is established, and the performance indicators (such as response delay, IOPS, throughput, etc.) are detected to determine whether storage performance problems occur. Then, a storage volume location algorithm based on clustering and multidimensional scoring model is proposed, which can locate the largest load storage volume causing storage performance problems. After long-term verification in many storage systems, the contention analysis framework proposed in this paper can identify and locate storage system performance problems efficiently and accurately, thus ensuring efficient utilization of cloud storage resources, and concurrently elevating the level of sustainable computing within cloud.