Abstract | ||
---|---|---|
•By “treating the computer as a network”, we extend the Stochastic Network Calculus (SNC)-based network latency analysis model to evaluate cloud storage service access latency, which consists of not only transfer latency on network, but also the CPU latency and disk read/write latency inside storage servers.•We propose a new metric, called resource-productivity, to describe the resource usage effectiveness of cloud storage systems. It is formally defined through a link function which combines tail latency and resource utilization.•We implement SMEA for cloud storage systems. It first adopts a Markov-modulated Poisson process (MMPP) to properly characterize the burstiness of workloads. By deriving statistical characterizations of each tenant’s trace from such basic model, it then integrate two predictors to accurately evaluate tail latency and resource utilization, and further implements a resource-productivity calculator. |
Year | DOI | Venue |
---|---|---|
2019 | 10.1016/j.sysarc.2019.04.002 | Journal of Systems Architecture |
Keywords | Field | DocType |
Cloud storage,Performance modeling,Stochastic network calculus,Resource productivity,Tail latency | Computer science,Latency (engineering),Testbed,Real-time computing,Burstiness,Queueing theory,Network calculus,Stochastic modelling,Cloud storage,Approximation error,Distributed computing | Journal |
Volume | ISSN | Citations |
98 | 1383-7621 | 0 |
PageRank | References | Authors |
0.34 | 0 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Binlei Cai | 1 | 3 | 2.43 |
Laipin Zhao | 2 | 0 | 0.34 |
Xiaobo Zhou | 3 | 64 | 16.25 |
Rongqi Zhang | 4 | 3 | 2.09 |
Keqiu Li | 5 | 1415 | 162.02 |