A new approach in the analysis and modeling of disk access patterns

Reliable performance evaluations require the use of representative workloads. This is no easy task since modern computer systems and their workloads are complex, with many interrelated attributes and complicated structures. Experts often use sophisticated mathematics to analyze and describe workload models, making these models difficult for practitioners to grasp. This book aims to close this gap by emphasizing the intuition and the reasoning behind the definitions and derivations related to the workload models. It provides numerous examples from real production systems, with hundreds of graphs. Using this book, readers will be able to analyze collected workload data and clean it if necessary, derive statistical models that include skewed marginal distributions and correlations, and consider the need for generative models and feedback from the system. The descriptive statistics techniques covered are also useful for other domains.

show abstract

“…• Self-similarity in access patterns [303,292,293,396]. A possible model for file access patterns is based on a hidden Markov model (HMM) [556].…”

Section: File System Access Patternsmentioning

confidence: 99%

Workload Modeling for Computer Systems Performance Evaluation

Feitelson

2015

196

131

View full text Add to dashboard Cite

show abstract

“…The literature describes and evaluates many techniques for generating synthetic block-level I/O workloads [11,12,13,15,16,24,25], file-level workloads [1,6,14,22], and application-level I/Os workloads [18]. File-and application-level synthesis techniques are important because they can be used to produce block-level workloads.…”

Section: Related Workmentioning

confidence: 99%

“…Jump distance within state calculates the jump distance between the beginning of the current request, and the end of the most recent request corresponding to the same state as the current request. For example, consider the following sequence of locations: 10,11,12,20,21,13,22,14,23,24,15. Assume each request size is 1 unit, and define states [10,19] and [20,29].…”

Section: Attributes and Attribute-valuesmentioning

confidence: 99%

See 1 more Smart Citation

Synthesizing representative I/O workloads using iterative distillation

Kurmas

Keeton

MacKenzie

11th IEEE/ACM International Symposium on Modeling, Analysis and Simulation of Computer Telecommunications Systems, 2003. MASCOT

View full text Add to dashboard Cite

show abstract

“…The Distiller is unique because, instead of attempting to preserve a set of attributes chosen a priori, it searches through a set of previously developed and studied attributes and generation techniques (e.g., [4,5,6,7,8,13,14,16]) then chooses those attributes that are most appropriate for the target workload and storage system under test. These existing techniques serve as the Distiller's library.…”

Section: Related Workmentioning

confidence: 99%