Fine-grained management of I/O optimizations based on workload characteristics

Abstract

With the advent of new computing paradigms, parallel file systems serve not only traditional scientific computing applications but also non-scientific computing applications, such as financial computing, business, and public administration. Parallel file systems provide storage services for multiple applications. As a result, various requirements need to be met. However, parallel file systems usually provide a unified storage solution, which cannot meet specific application needs. In this paper, an extended file handle scheme is proposed to deal with this problem. The original file handle is extended to record I/O optimization information, which allows file systems to specify optimizations for a file or directory based on workload characteristics. Therefore, fine-grained management of I/O optimizations can be achieved. On the basis of the extended file handle scheme, data prefetching and small file optimization mechanisms are proposed for parallel file systems. The experimental results show that the proposed approach improves the aggregate throughput of the overall system by up to 189.75%.

Keywords

parallel file systems / workload characteristics / extended file handle / data prefetching / small files

References

[1] Carns P H, Ligon W B, Ross R B, Thakur R. PVFS: a parallel file system for Linux clusters. In: Proceedings of the 4th Annual Linux Showcase and Conference. 2000, 317–327

[2] Schmuck F, Haskin R. GPFS: a shared-disk file system for large computing clusters. In: Proceedings of the 10th USENIX Conference on File and Storage Technologies. 2002, 231–244

[3] Wei B, Xiao L, Zhou B, Qin G, Yan B, Huo Z. I/O optimizations based on workload characteristics for parallel file systems. In: Proceedings of the 16th Annual IFIP International Conference on Network and Parallel Computing. 2019, 305–310

[4] Isaila F, Balaprakash P, Wild S M, Kimpe D, Latham R, Ross R, Hovland P. Collective I/O tuning using analytical and machine learning models. In: Proceedings of the IEEE International Conference on Cluster Computing. 2015, 128–137

[5] Byna S, Chen Y, Sun X H, Thakur R, Gropp W. Parallel I/O prefetching using MPI file caching and I/O signatures. In: Proceedings of the 2008 ACM/IEEE Conference on Supercomputing. 2008, 1–12

[6] Chen J, Liu J, Roth P, Chen Y. Using working set reorganization to manage storage systems with hard and solid state disks. In: Proceedings of the 43rd International Conference on Parallel Processing Workshops. 2014, 283–291

[7] Costa L B, Ripeanu M. Towards automating the configuration of a distributed storage system. In: Proceedings of the 11th IEEE/ACM International Conference on Grid Computing. 2010, 201–208

[8] Narayan S, Chandy J. Attest: attributes-based extendable storage. Journal of Systems and Software, 2010, 83(4): 548–556

[9] Madhyastha T M, Reed D A. Learning to classify parallel input/output access patterns. IEEE Transactions on Parallel and Distributed Systems, 2002, 13(8): 802–813

[10] Wang Y, Kaeli D. Profile-guided I/O partitioning. In: Proceedings of the 17th Annual International Conference on Supercomputing. 2003, 252–260

[11] Habermann P, Chi C C, Alvarez-Mesa M, Juurlink B. Application-specific cache and prefetching for HEVC CABAC decoding. IEEE MultiMedia, 2017, 24(1): 72–85

[12] Chen J, Roth P C, Chen Y. Using pattern-models to guide SSD deployment for big data applications in HPC systems. In: Proceedings of IEEE International Conference on Big Data. 2013, 332–337

[13] He J, Bent J, Torres A, Grider G, Gibson G, Maltzahn C, Sun X H. I/O acceleration with pattern detection. In: Proceedings of the 22nd International Symposium on High-Performance Parallel and Distributed Computing. 2013, 25–36

[14] Patrick C M, Kandemir M, Karakoy M, Son S W, Choudhary A. Cashing in on hints for better prefetching and caching in PVFS and MPI-IO. In: Proceedings of the 19th ACM International Symposium on High Performance Distributed Computing. 2010, 191–202

[15] Battle L, Chang R, Stonebraker M. Dynamic prefetching of data tiles for interactive visualization. In: Proceedings of the 2016 International Conference on Management of Data. 2016, 1363–1375

[16] Al-Kiswany S, Gharaibeh A, Ripeanu M. The case for a versatile storage system. ACM SIGOPS Operating Systems Review, 2010, 44(1): 10–14

[17] Calderon A, Garcia-Carballeira F, Sanchez L M, Garcia J D, Fernandez J. Fault tolerant file models for parallel file systems: introducing distribution patterns for every file. The Journal of Supercomputing, 2009, 47(3): 312–334

[18] Qiu M, Sha E H M. Cost minimization while satisfying hard/soft timing constraints for heterogeneous embedded systems. ACM Transactions on Design Automation of Electronic Systems, 2009, 14(2): 1–30

[19] Vilayannur M, Nath P, Sivasubramaniam A. Providing tunable consistency for a parallel file store. In: Proceedings of the 4th USENIX Conference on File and Storage Technologies. 2005, 17–30

[20] Xue J, Yan F, Birke R, Chen L Y, Scherer T, Smirni E. PRACTISE: robust prediction of data center time series. In: Proceedings of the 11th International Conference on Network and Service Management. 2015, 126–134

[21] Dai D, Bao F S, Zhou J, Chen Y. Block2vec: a deep learning strategy on mining block correlations in storage systems. In: Proceedings of the 45th International Conference on Parallel Processing Workshops. 2016, 230–239

[22] Guo C, Li Y, Liu H, Wu Z. An application-oriented cache allocation and prefetching method for long-running applications in distributed storage systems. Chinese Journal of Electronics, 2019, 28(4): 773–780

[23] Zhang S L, Catanese H, Wang A A I. The composite-file file system: decoupling the one-to-one mapping of files and metadata for better performance. In: Proceedings of the 14th USENIX Conference on File and Storage Technologies. 2016, 15–22

[24] Hou B, Chen F. Pacaca: mining object correlations and parallelism for enhancing user experience with cloud storage. In: Proceedings of the 26th IEEE International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems. 2018, 293–305

[25] Sheoran S, Sethia D, Saran H. Optimized mapfile based storage of small files in hadoop. In: Proceedings of the 17th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing. 2017, 906–912

[26] Mehmood A, Usman M, Mehmood W, Khaliq Y. Performance efficiency in hadoop for storing and accessing small files. In: Proceedings of the 7th International Conference on Innovative Computing Technology. 2017, 211–216

[27] Carns P, Lang S, Ross R, Vilayannur M, Kunkel J, Ludwig T. Small-file access in parallel file systems. In: Proceedings of the IEEE International Symposium on Parallel & Distributed Processing. 2009, 1–11

[28] Kuhn M, Kunkel JM, Ludwig T. Dynamic file system semantics to enable metadata optimizations in PVFS. Concurrency and Computation: Practice and Experience, 2009, 21(14): 1775–1788

[29] Wei B, Xiao L M, Wei W, Song Y, Zhou B Y. A new adaptive coding selection method for distributed storage systems. IEEE Access, 2018, 6(1): 13350–13357

[30] Li Z P, Yu H, Liu Y C, Liu F Q. An improved adaptive exponential smoothing model for short-term travel time forecasting of urban arterial street. Acta Automatica Sinica, 2008, 34(11): 1404–1409

[31] Weil S A, Brandt S A, Miller E L, Long D D E. Ceph: a scalable, highperformance distributed file system. In: Proceedings of the 7th USENIX Symposium on Operating Systems Design and Implementation. 2006, 307–320

[32] Shvachko K, Kuang H, Radia S, Chansler R. The hadoop distributed file system. In: Proceedings of the 26th IEEE Symposium on Mass Storage Systems and Technologies. 2010, 1–10

[33] Ghemawat S, Gobioff H, Leung S T. The Google file system. In: Proceedings of the 19th ACM Symposium on Operating Systems Principles. 2003, 29–43

RIGHTS & PERMISSIONS

Higher Education Press