Abstract
The ever-increasing gap between compute and I/O performance in HPC platforms, together with the development of novel NVMe storage devices (NVRAM), led to the emergence of the burst buffer concept—an intermediate persistent storage layer logically positioned between random-access main memory and a parallel file system. Despite the development of real-world architectures as well as research concepts, resource and job management systems, such as Slurm, provide only marginal support for scheduling jobs with burst buffer requirements, in particular ignoring burst buffers when backfilling. We investigate the impact of burst buffer reservations on the overall efficiency of online job scheduling for common algorithms: First-Come-First-Served (FCFS) and Shortest-Job-First (SJF) EASY-backfilling. We evaluate the algorithms in a detailed simulation with I/O side effects. Our results indicate that the lack of burst buffer reservations in backfilling may significantly deteriorate scheduling. We also show that these algorithms can be easily extended to support burst buffers. Finally, we propose a burst-buffer–aware plan-based scheduling algorithm with simulated annealing optimisation, which improves the mean waiting time by over 20% and mean bounded slowdown by 27% compared to the burst-buffer–aware SJF-EASY-backfilling.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
References
Ben-Ameur, W.: Computing the initial temperature of simulated annealing. Comput. Optim. Appl. 29, 369–385 (2004). https://doi.org/10.1023/B:COAP.0000044187.23143.bd
Bhimji, W., et al.: Accelerating science with the NERSC burst buffer early user program (2016)
Carastan-Santos, D., De Camargo, R.Y., Trystram, D., Zrigui, S.: One can only gain by replacing EASY Backfilling: a simple scheduling policies case study. In: CCGrid. IEEE (2019)
Casanova, H., Giersch, A., Legrand, A., Quinson, M., Suter, F.: Versatile, scalable, and accurate simulation of distributed applications and platforms. J. Parallel Distrib. Comput. 74, 2899–2917 (2014)
Dongarra, J.: Report on the Fujitsu Fugaku system. Tech. Rep. ICL-UT-20-06, University of Tennessee, June 2020
Dutot, P.F., Mercier, M., Poquet, M., Richard, O.: Batsim: a realistic language-independent resources and jobs management systems simulator. In: JSSPP Workshop (2016)
Fan, Y., et al.: Scheduling beyond CPUs for HPC. In: HPDC 2019. ACM (2019)
Feitelson, D.G.: Experimental analysis of the root causes of performance evaluation results: a backfilling case study. TPDS 16, 175–182 (2005)
Feitelson, D.G., Tsafrir, D., Krakov, D.: Experience with using the parallel workloads archive. J. Parallel Distrib. Comput. 74, 1982–2967 (2014)
Gainaru, A., Aupy, G., Benoit, A., Cappello, F., Robert, Y., Snir, M.: Scheduling the I/O of HPC applications under congestion. In: IPDPS, Proceedings IEEE (2015)
Harms, K., Oral, H.S., Atchley, S., Vazhkudai, S.S.: Impact of burst buffer architectures on application portability (2016)
Hemmert, K.S., et al.: Trinity: architecture and early experience (2016)
Herbein, S., et al.: Scalable I/O-aware job scheduling for burst buffer enabled HPC clusters. In: HPDC 2016. ACM (2016)
Hofmann, H., Wickham, H., Kafadar, K.: Letter-value plots: boxplots for large data. J. Comput. Graph. Stat. 26, 469–477 (2017)
Hovestadt, M., Kao, O., Keller, A., Streit, A.: Scheduling in HPC resource management systems: queuing vs. planning. In: Feitelson, D., Rudolph, L., Schwiegelshohn, U. (eds.) JSSPP 2003. LNCS, vol. 2862, pp. 1–20. Springer, Heidelberg (2003). https://doi.org/10.1007/10968987_1
Isakov, M., et al.: HPC I/O throughput bottleneck analysis with explainable local models. In: SC20. IEEE (2020)
Klusáček, D., Tóth, Š, Podolníková, G.: Real-life experience with major reconfiguration of job scheduling system. In: Desai, N., Cirne, W. (eds.) JSSPP, pp. 83–101. Springer International Publishing, Cham (2017). https://doi.org/10.1007/978-3-319-61756-5_5
Kopanski, J., Rzadca, K.: Artifact and instructions to generate experimental results for the Euro-par 2021 paper: plan-based job scheduling for supercomputers with shared burst buffers, August 2021. https://doi.org/10.6084/m9.figshare.14754507
Lackner, L.E., Fard, H.M., Wolf, F.: Efficient job scheduling for clusters with shared tiered storage. In: CCGRID, IEEE/ACM, pp. 321–330 (2019)
Liu, N., et al.: On the role of burst buffers in leadership-class storage systems. In: MSST, Proceedings IEEE (2012)
Poquet, M.: Simulation approach for resource management. Theses, Université Grenoble Alpes, December 2017
RIKEN Center for Computational Science: Post-k (fugaku) information (2020). https://postk-web.r-ccs.riken.jp/spec.html. Accessed 04 Aug 2020
Srinivasan, S., Kettimuthu, R., Subramani, V., Sadayappan, P.: Selective reservation strategies for backfill job scheduling. In: Feitelson, D.G., Rudolph, L., Schwiegelshohn, U. (eds.) JSSPP 2002. LNCS, vol. 2537, pp. 55–71. Springer, Heidelberg (2002). https://doi.org/10.1007/3-540-36180-4_4
Vazhkudai, S.S., et al.: The design, deployment, and evaluation of the coral pre-exascale systems. In: SC18, Proceedings IEEE (2018)
Yoo, A.B., Jette, M.A., Grondona, M.: SLURM: simple Linux utility for resource management. In: Feitelson, D., Rudolph, L., Schwiegelshohn, U. (eds.) JSSPP 2003. LNCS, vol. 2862, pp. 44–60. Springer, Heidelberg (2003). https://doi.org/10.1007/10968987_3
Zheng, X., Zhou, Z., Yang, X., Lan, Z., Wang, J.: Exploring plan-based scheduling for large-scale computing systems. In: CLUSTER, Proceedings IEEE (2016)
Zhou, Z., et al.: I/O-aware batch scheduling for petascale computing systems. In: CLUSTER. IEEE (2015)
Acknowledgements
This research is supported by a Polish National Science Center grant Opus (UMO-2017/25/B/ST6/00116).
The MetaCentrum workload log [17] was graciously provided by Czech National Grid Infrastructure MetaCentrum. The workload log from the KTH SP2 was graciously provided by Lars Malinowsky.
Author information
Authors and Affiliations
Corresponding authors
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2021 Springer Nature Switzerland AG
About this paper
Cite this paper
Kopanski, J., Rzadca, K. (2021). Plan-Based Job Scheduling for Supercomputers with Shared Burst Buffers. In: Sousa, L., Roma, N., Tomás, P. (eds) Euro-Par 2021: Parallel Processing. Euro-Par 2021. Lecture Notes in Computer Science(), vol 12820. Springer, Cham. https://doi.org/10.1007/978-3-030-85665-6_8
Download citation
DOI: https://doi.org/10.1007/978-3-030-85665-6_8
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-85664-9
Online ISBN: 978-3-030-85665-6
eBook Packages: Computer ScienceComputer Science (R0)