SLA-Aware Adaptive Mapping Scheme in Bigdata Distributed Storage Systems
Sopanhapich Chum, Jongmoo Choi, Jerry Li, Heekwon Park · 2020
As data are processed by diverse clients ranging from urgent time-critical to best-effort, supporting different QoS (Quality of Service) becomes a vital component in a distributed storage system. In this paper, we propose a novel SLA (Service Level Agreement)-aware adaptive mapping scheme that can differentiate between urgent and normal clients based on their I/O requirements. The scheme basically divides storage into two regions, normal and urgent, which makes it feasible to isolate urgent clients from normal ones. In addition, it changes the size of the isolated region in an adaptive manner so that it can provide better performance to normal clients. Besides, we devise two techniques, called logical cluster and normal inclusion, to prevent interference caused by adaptability. We implement our scheme using the CRUSH (Controlled Replication Under Scalable Hashing) mapping algorithm in Ceph, a well-known distributed storage system. Evaluation results demonstrate that our proposal can comply with SLA requirements while providing better performance than the fixed mapping scheme.