Lunule: An Agile and Judicious Metadata Load Balancer for CephFS
SessionFile System
Event Type
Paper
Cloud and Distributed Computing
File Systems and I/O
Storage
TP
TimeWednesday, 17 November 202111am - 11:30am CST
Location227-228
DescriptionTo scale out the massive metadata access, the Ceph file system (CephFS) adopts a dynamic subtree partitioning method, splitting the hierarchical namespace and distributing subtrees across multiple metadata servers. This method suffers a severe imbalance problem, however; resulting in poor performance mainly because of its inaccurate imbalance prediction, ignorance of workload characteristics and unnecessary/invalid migration activities. To eliminate these inefficiencies, we propose Lunule, a novel CephFS metadata load balancer, which employs an imbalance factor model for accurately determining when to trigger re-balance and tolerating benign imbalanced situations, and a workload-aware migration planner to appropriately select subtree migration candidates. Compared to state-of-the-art baselines, Lunule achieves better load balance, increases the metadata throughput by up to 315.8% and shortens the tail job completion time by up to 64.6% for four real-world workloads and their mixture, respectively. Additionally, Lunule is capable of handling the metadata cluster expansion and the client workload growth.
Download PDF
Archive view
Authors
