A Reproducible I/O Analyzer

  • Rodrigo P. do Nascimento IPT
  • Alfredo Goldman Vel Lejbman USP

Resumo


This paper presents the Stealth Metrics Framework, an I/O characterization and data-driven I/O replay framework that users can utilize to gain detailed insights into how their applications issue I/O to storage systems. The study provides a clear snapshot of how a MongoDB instance manages recovery from a crash from a storage I/O perspective, and explores potential strategies to enhance recovery performance.

Referências

Desnoyers, M. (2020). Using the linux kernel tracepoints. Accessed on: 10 April 2024.

Gebai, M. and Dagenais, M. R. (2018). Survey and analysis of kernel and userspace tracers on linux: Design, implementation, and overhead. ACM Computing Surveys, 51(2):1–33.

Kerrisk, M. (2010). The Linux Programming Interface: A Linux and UNIX System Programming Handbook. No Starch Press.

MongoDB (2024). Production notes - mongodb manual. Accessed on: 11 April 2024.

Reisel, D., Gantz, J., and Rydning, J. (2020). The digitalization of the world from edge to core. White paper, International Data Corporation.

Wright, S. A. (2019). Performance modeling, benchmarking and simulation of high performance computing systems. Future Generation Computer Systems, 92:900–902.
Publicado
16/05/2024
NASCIMENTO, Rodrigo P. do; VEL LEJBMAN, Alfredo Goldman. A Reproducible I/O Analyzer. In: ESCOLA REGIONAL DE ALTO DESEMPENHO DE SÃO PAULO (ERAD-SP), 15. , 2024, Rio Claro/SP. Anais [...]. Porto Alegre: Sociedade Brasileira de Computação, 2024 . p. 97-100. DOI: https://doi.org/10.5753/eradsp.2024.239940.