2011 · 102 citations · 21 references
ComputingComputer ScienceComputer SystemsComputer ArchitectureScientific ComputingStorage SystemsData Storage SystemsFile SystemsScalable Data StorageStorage EnginesStorage ManagementApplication Behavior MonitoringApplication-level PerformanceFile System Activity AnalysisFile System PerformanceFile System Performance EvaluationStorage Access PatternsStorage Performance ModelingStorage System Simulation
Computational science drives a growing need for powerful storage, yet continuous characterization of large‑scale production systems remains difficult due to their size and complexity. This study proposes a scalable, continuous, system‑wide I/O characterization methodology to better understand the diverse workloads that such storage systems must support. The approach integrates storage‑device instrumentation, static file‑system analysis, and a novel mechanism for capturing detailed application‑level behavior across millions of jobs and petabytes of traffic. Applying the method to a two‑month study of a 557‑teraflop IBM Blue Gene/P system revealed system‑wide trends, identified prolific storage users, correlated I/O strategies with performance, and enabled tuning of applications, thereby demonstrating the method’s effectiveness and informing future storage‑system design.
Computational science applications are driving a demand for increasingly powerful storage systems. While many techniques are available for capturing the I/O behavior of individual application trial runs and specific components of the storage system, continuous characterization of a production system remains a daunting challenge for systems with hundreds of thousands of compute cores and multiple petabytes of storage. As a result, these storage systems are often designed without a clear understanding of the diverse computational science workloads they will support.In this study, we outline a methodology for scalable, con tinuous, systemwide I/O characterization that combines storage device instrumentation, static file system analysis, and a new mechanism for capturing detailed application-level behavior. This methodology allows us to quantify systemwide trends such as the way application behavior changes with job size, the "burstiness" of the storage system, and the evolution of file system contents over time. The data also can be examined by application domain to determine the most prolific storage users and also investigate how their I/O strategies correlate with I/O performance. At the most detailed level, our characterization methodology can also be used to focus on individual applications and guide tuning efforts for those applications. We demonstrate the effectiveness of our methodology by performing a multilevel, two-month study of Intrepid, a 557 teraflop IBM Blue Gene/P system. During that time, we captured application-level I/O characterizations from 6,481 unique jobs spanning 38 science and engineering projects with up to 163,840 processes per job. We also captured patterns of I/O activity in over 8 petabytes of block device traffic and summarized the con tents of file systems containing over 191 million flies. We then used the results of our study to tune example applications, highlight trends that impact the design of future storage systems, and identify opportunities for improvement in I/O characterization methodology.
21
GPFS: A Shared-Disk File System for Large Computing Clusters
Frank Schmuck, Roger Haskin · 2002 · 1.2K citations
Scalable performance analysis: the Pablo performance analysis environment
Daniel A. Reed, Philip C. Roth, Ruth Aydt et al. · 2002 · 258 citations
Measurement and analysis of large-scale network file system workloads
Andrew W. Leung, Shankar Pasupathy, Garth R. Goodson et al. · 2008 · 221 citations
24/7 Characterization of petascale I/O workloads
Philip Carns, Robert Latham, Robert Ross et al. · 2009 · 212 citations
A nine year study of file system and storage benchmarking
Avishay Traeger, Erez Zadok, Nikolai Joukov et al. · ACM Transactions on Storage · 2008 · 205 citations