Carnegie Mellon University Parallel Data Lab Technical Report CMU-PDL-22-102, February 2022. Superceded by SYSTOR '23: Proceedings of the 16th ACM International Conference on Systems and Storage, Haifa, Israel, June 5-7, 2023.
Hojin Park, Gregory R. Ganger, George Amvrosiadis
Carnegie Mellon University
Public cloud providers offer a diverse collection of block storage options with different costs and performance SLAs. As a consequence, it is difficult to select the right allocations for storage backends when moving data-heavy applications to the cloud. Mimir is a tool for automatically finding a cost-efficient virtual storage cluster (VSC) configuration for a customer’s storage workload and performance requirements. Importantly, since no single allocation type is best for all workloads, Mimir considers all allocation types and even heterogeneous mixes of them. In our experiments, compared to state-of-the-art approaches that consider only one allocation type, Mimir finds VSC configurations that reduce cost by up to 81% for substantial storage workloads.
KEYWORDS: Public cloud, Storage system, Automated resource provisioning
FULL TR: pdf
FULL CONFERENCE PAPER: pdf