Abstract
We introduce new methods to replay intensive block I/O workloads more accurately. These methods can be used to reproduce realistic workloads for benchmarking, performance validation, and tuning of a high-performance block storage device/system. In this article, we study several sources in the stock operating system that introduce uncertainty in the workload replay. Based on the remedies of these findings, we design and develop a new replay tool called hfplayer that replays intensive block I/O workloads in a similar unscaled environment with more accuracy. To replay a given workload trace in a scaled environment with faster storage or host server, the dependency between I/O requests becomes crucial since the timing and ordering of I/O requests is expected to change according to these dependencies. Therefore, we propose a heuristic way of speculating I/O dependencies in a block I/O trace. Using the generated dependency graph, hfplayer tries to propagate I/O related performance gains appropriately along the I/O dependency chains and mimics the original application behavior when it executes in a scaled environment with slower or faster storage system and servers. We evaluate hfplayer with a wide range of workloads using several accuracy metrics and find that it produces better accuracy when compared to other replay approaches.
- Percona. 2012. Replaying Database Load With Percona Playback. Retrieved October 16, 2017, from http://www.percona.com/blog/2012/10/25/replaying-database-load-with-percona-playback/.Google Scholar
- Spinics.net. 2014. Re: fio replay. Retrieved October 16, 2017, from http://www.spinics.net/lists/fio/msg02817.html.Google Scholar
- HammerDB. 2017. Home Page. Retrieved October 16, 2017, from http://www.hammerdb.com.Google Scholar
- SPDK. 2017. Storage Performance Development Kit. Retrieved October 16, 2017, from http://www.spdk.io.Google Scholar
- Alireza Haghdoost. 2017. Linux Bug Fix Patch: Enable Sysfs Nomerge Control for I/O Requests in the Plug List. Retrieved October 16, 2017, from https://lkml.org/lkml/2013/10/22/541.Google Scholar
- Percona. 2017. Download Percona Lab Query Playback. Retrieved October 16, 2017, from http://www.percona.com/doc/percona-playback.Google Scholar
- SNIA. 2017. IOTTA Repository Home. Retrieved October 16, 2017, from http://iotta.snia.org.Google Scholar
- Storage Performance Council. 2017. SPC-1 Benchmark Results for Top 10 Block Storage by Price-Performance. Available at http://www.storageperformance.org.Google Scholar
- Eric Anderson. 2008. Buttress, a cautionary tale. In Proceedings of the 2008 File and Storage Systems Benchmarking Workshop.Google Scholar
- Eric Anderson, Mahesh Kallahalla, Mustafa Uysal, and Ram Swaminathan. 2004. Buttress: A toolkit for flexible and high fidelity I/O benchmarking. In Proceedings of the 3rd USENIX Conference on File and Storage Technologies. Google Scholar
Digital Library
- Matias Bjørling, Jens Axboe, David Nellans, and Philippe Bonnet. 2013. Linux block IO: Introducing multi-queue SSD access on multi-core systems. In Proceedings of the 6th International Systems and Storage Conference. ACM, New York, NY, 22. Google Scholar
Digital Library
- Dimitri Bourilkov, Paul Ralph Avery, Yu Fu, Craig Prescott, and Bockjoo Kim. The lustre filesystem for petabyte storage at the Florida HPC center. In Proceedings of the HEPiX Fall 2014 Workshop.Google Scholar
- Alan D. Brunelle. 2011. Btrecord and btreplay user guide. Retrieved October 16, 2017, from http://www.fis.unipr.it/doc/blktrace-1.0.1/btreplay.pdf.Google Scholar
- Alireza Haghdoost, Weiping He, Jerry Fredin, and David H. C. Du. 2017. On the accuracy and scalability of intensive I/O workload replay. In Proceedings of the 15th USENIX Conference on File and Storage Technologies (FAST’17). 315--328. https://www.usenix.org/conference/fast17/technical-sessions/presentation/haghdoost. Google Scholar
Digital Library
- Nikolai Joukov, Timothy Wong, and Erez Zadok. 2005. Accurate and efficient replaying of file system traces. In Proceedings of the 4th USENIX Conference on File and Storage Technologies (FAST’05), Vol. 4. 25. Google Scholar
Digital Library
- Zhuo Liu, Fei Wu, Xiao Qin, Changsheng Xie, Jian Zhou, and Jianzong Wang. 2010. TRACER: A trace replay tool to evaluate energy-efficiency of mass storage systems. In Proceedings of the 2010 IEEE International Conference on Cluster Computing. 68--77. Google Scholar
Digital Library
- Michael P. Mesnier, Matthew Wachs, Raja R. Simbasivan, Julio Lopez, James Hendricks, Gregory R. Ganger, and David R. O’Hallaron. 2007. //Trace: Parallel trace replay with approximate causal events. In Proceedings of the 5th USENIX Conference on File and Storage Technologies (FAST’07). 24. Google Scholar
Digital Library
- Al Morton, Len Ciavattone, Gomathi Ramachandran, Stanislav Shalunov, and Jerry Perser. 2006. Packet Reordering Metrics. IETF Internet-standard: RFC4737.Google Scholar
- Thiago Emmanuel Pereira, Francisco Brasileiro, and Lívia Sampaio. 2016. File system trace replay methods through the lens of metrology. In Proceedings of the 32nd International Conference on Massive Storage Systems and Technology (MSST’16).Google Scholar
Cross Ref
- Bianca Schroeder, Adam Wierman, and Mor Harchol-Balter. 2006. Open versus closed: A cautionary tale. In Proceedings of the 3rd Conference on Networked Systems Design and Implementation (NSDI’06), Vol. 6. 18. Google Scholar
Digital Library
- Thomas Schbel-Theuer. 2012. blkreplay and Sonar Diagrams. Available at http://www.blkreplay.org.Google Scholar
- Sankaran Sivathanu, Jinpyo Kim, Devaki Kulkarni, and Ling Liu. 2011. Load-aware replay of I/O traces. In Proceedings of the 9th USENIX Conference on File and Storage Technologies (FAST’11), Work in Progress (WiP) Session.Google Scholar
- Storage Networking Industry Association (SNIA). 2010. Block I/O trace common semantics. Working draft.Google Scholar
- Vasily Tarasov. 2013. Multi-Dimensional Workload Analysis and Synthesis for Modern Storage Systems. Ph.D. Dissertation. Stony Brook University. Google Scholar
Digital Library
- Vasily Tarasov, Santhosh Kumar, Jack Ma, Dean Hildebrand, Anna Povzner, Geoff Kuenning, and Erez Zadok. 2012. Extracting flexible, replayable models from large block traces. In Proceedings of the 10th USENIX Conference on File and Storage Technologies (FAST’12). 22. Google Scholar
Digital Library
- Vasily Tarasov, Santhosh Kumar, Jack Ma, Dean Hildebrand, Anna Povzner, Geoff Kuenning, and Erez Zadok. 2012. Extracting flexible, replayable models from large block traces. In Proceedings of the 10th USENIX Conference on File and Storage Technologies (FAST’12), Vol. 12. 22. Google Scholar
Digital Library
- Vasily Tarasov, Erez Zadok, and Spencer Shepler. 2016. Filebench: A flexible framework for file system benchmarking. The USENIX Magazine 41, 1, 6--12.Google Scholar
- Mojtaba Tarihi, Hossein Asadi, and Hamid Sarbazi-Azad. 2015. DiskAccel: Accelerating disk-based experiments by representative sampling. ACM SIGMETRICS Performance Evaluation Review 43, 1, 297--308. Google Scholar
Digital Library
- Avishay Traeger, Erez Zadok, Nikolai Joukov, and Charles P. Wright. 2008. A nine year study of file system and storage benchmarking. ACM Transactions on Storage 4, 2, 5. Google Scholar
Digital Library
- R. Weber. 2005. SCSI architecture model-3 (SAM-3). ANSI INCITS 402.Google Scholar
- Zev Weiss, Tyler Harter, Andrea C. Arpaci-Dusseau, and Remzi H. Arpaci-Dusseau. 2013. ROOT: Replaying multithreaded traces with resource-oriented ordering. In Proceedings of the 24th ACM Symposium on Operating Systems Principles (SOSP’13). ACM, New York, NY, 373--387. Google Scholar
Digital Library
- Ningning Zhu, Jiawu Chen, Tzi-Cker Chiueh, and Daniel Ellard. 2005. TBBT: Scalable and accurate trace replay for file server evaluation. ACM SIGMETRICS Performance Evaluation Review 33, 392--393. Google Scholar
Digital Library
Index Terms
hfplayer: Scalable Replay for Intensive Block I/O Workloads
Recommendations
Taming IO Spikes in Enterprise and Campus VM Deployment
SYSTOR 2014: Proceedings of International Conference on Systems and StorageEnterprises and campuses have widely employed virtualization to improve resource utilization and save capital costs. The virtualized storage infrastructure usually comprises a pool of storage nodes (disks, RAID groups and storage servers), each ...
Design and Implementation of an Out-of-Band Virtualization System for Large SANs
Out-of-band virtualization intrinsically has the potential to provide high performance and good scalability. Unfortunately, existing out-of-band virtualization systems have some limitations such as restrictions to specific platforms and/or hardware. In ...
Raid 5 for Secured Storage Virtualisation
DSDE '10: Proceedings of the 2010 International Conference on Data Storage and Data EngineeringVirtualization is an important step towards improving overall IT efficiency and ROI. Implementing storage virtualization, in other words, extends the benefits of an investment in server virtualization with simplified storage management, improved storage ...






Comments