Abstract: Emerging non-volatile memory technologies, especially flash-based solid state drives (SSDs), have increasingly been adopted in the storage stack. They provide numerous advantages over traditional mechanically rotating hard disk drives (HDDs) and have a tendency to replace HDDs. Due to the long existence of HDDs as primary building blocks for storage systems, however, much of the system software has been specially designed for HDD and may not be optimal for non-volatile memory media. Therefore, in order to realistically leverage its superior raw performance to the maximum, the existing upper layer software has to be re-evaluated or re-designed. To this end, in this paper, we propose PASS, an optimized i/O scheduler at the Linux block layer to accommodate the changing trend of underlying storage devices toward flash-based SSDs. PASS takes the rich internal parallelism in SSDs into account when dispatching requests to the device driver in order to achieve high performance. Specifically, it partitions the logical storage space into fixed-size regions (preferably the component package sizes) as scheduling units. These scheduling units are serviced in a round-robin manner and for every chance that the chosen dispatching unit issues only a batch of either read or write requests to suppress the excessive mutual interference. Additionally, the requests are sorted according to their visiting addresses while waiting in the dispatching queues to exploit high sequential performance of SSD. The experimental results with a variety of workloads have shown that PASS outperforms the four Linux off-the-shelf i/O schedulers by a degree of 3% up to 41%, while at the same time it improves the lifetime significantly, due to reducing the internal write amplification.

PASS：一种简单、高效、并行感知的固态盘I/O调度器

[1]Agrawal, N., Prabhakaran, V., Wobber, T., et al., 2008. Design tradeoffs for SSD performance. Proc. USENIX Annual Technical Conf., p.57-70.

[2]Andersen, D., Franklin, J., Kaminsky, M., et al., 2009. Fawn: a fast array of wimpy nodes. Proc. 22nd ACM Symp. on Operating Systems Principles, p.1-14.

[3]Badam, A., Pai, V.S., 2011. SSDAlloc: hybrid SSD/RAM memory management made easy. Proc. 8th USENIX Symp. on Networked Systems Design and Implementation, p.16.

[4]Bhadkamkar, M., Guerra, J., Useche, L., et al., 2009. BORG: block-reorganization for self-optimizing storage systems. Proc. 7th Conf. on File and Storage Technologies, p.183-196.

[5]Boboila, S., Desnoyers, P., 2010. Write endurance in flash drives: measurements and analysis. Proc. 8th USENIX Conf. on File and Storage Technologies, p.115-128.

[6]Caulfield, A.M., Grupp, L.M., Swanson, S., 2009. Gordon: using flash memory to build fast, power-efficient clusters for data-intensive applications. Proc. 14th Int. Conf. on Architectural Support for Programming Languages and Operating Systems, p.217-228.

[7]Caulfield, A.M., de Arup, Coburn, J., et al., 2010. Moneta: a high-performance storage array architecture for next-generation, non-volatile memories. Proc. 43rd Annual IEEE/ACM Int. Symp. on Microarchitecture, p.385-395.

[8]Caulfield, A.M., Mollov, T.I., Eisner, L.A., 2012. Providing safe, user space access to fast, solid state disks. Proc. 17th Int. Conf. on Architectural Support for Programming Languages and Operating Systems, p.387-400.

[9]Chen, F., Koufaty, D.A., Zhang, X., 2009. Understanding intrinsic characteristics and system implications of flash memory based solid state drives. Proc. 11th Int. Joint Conf. on Measurement and Modeling of Computer Systems, p.181-192.

[10]Chen, F., Lee, R., Zhang X., 2011a. Essential roles of exploiting internal parallelism of flash memory based solid state drives in high-speed data processing. Proc. 17th IEEE Int. Symp. on High Performance Computer Architecture, p.266-277.

[11]Chen, F., Luo, T., Zhang, X., 2011b. CAFTL: a content-aware flash translation layer enhancing the lifespan of flash memory based solid state drives. Proc. 9th USENIX Conf. on File and Storage Technologies, p.1-14.

[12]Dunn, M., Reddy, A.L.N., 2009. A New I/O Scheduler for Solid State Devices. PhD Thesis, Texas A&M University, Texas, USA.

[13]Gal, E., Toledo, S., 2005. Algorithms and data structures for flash memories. ACM Comput. Surv., 37(2):138-163.

[14]Grupp, L.M., Davis, J.D., Swanson, S., 2012. The bleak future of NAND flash memory. Proc. 10th USENIX Conf. on File and Storage Technologies, p.2.

[15]Gupta, A., Pisolkar, R., Urgaonkar, B., et al., 2011. Leveraging value locality in optimizing NAND flash-based SSDs. Proc. 9th USENIX Conf. on File and Storage Technologies, p.91-103.

[16]Hu, X.Y., Eleftheriou, E., Haas, R., et al., 2009. Write amplification analysis in flash-based solid state drives. Proc. SYSTOR: the Israeli Experimental Systems Conf., p.10.

[17]Hu, Y., Jiang, H., Feng, D., et al., 2011. Performance impact and interplay of SSD parallelism through advanced commands, allocation strategy and data granularity. Proc. 25th Int. Conf. on Supercomputing, p.96-107.

[18]Huang, H., Hung, W., Shin, K.G., 2005. FS2: dynamic data replication in free disk space for improving disk performance and energy consumption. Proc. 20th ACM Symp. on Operating Systems Principles, p.263-276.

[19]Iyer, S., Druschel, P., 2001. Anticipatory scheduling: a disk scheduling framework to overcome deceptive idleness in synchronous I/O. Proc. 18th ACM Symp. on Operating Systems Principles, p.117-130.

[20]Kim, J., Oh, Y., Kim, E., et al., 2009. Disk schedulers for solid state drives. Proc. 7th ACM Int. Conf. on Embedded Software, p.295-304.

[21]Koller, R., Rangaswami, R., 2010. I/O deduplication: utilizing content similarity to improve I/O performance. Proc. 8th USENIX Conf. on File and Storage Technologies, p.16.

[22]Liu, R.S., Yang, C.L., Wu, W., 2012. Optimizing NAND flash-based SSDs via retention relaxation. Proc. 10th USENIX Conf. on File and Storage Technologies, p.11.

[23]Lu, Y., Shu, J., Zheng, W., 2013. Extending the lifetime of flash-based storage through reducing write amplification from file systems. Proc. 11th USENIX Conf. on File and Storage Technologies, p.257-270.

[24]Mina, C., Kimb, K., Choc, H., et al., 2012. SFS: random write considered harmful in solid state drives. Proc. 10th USENIX Conf. on File and Storage Technologies, p.12.

[25]Park, S., Shen, K., 2012. FIOS: a fair, efficient flash I/O scheduler. Proc. 10th USENIX Conf. on File and Storage Technologies, p.13.

[26]Ren, J., Yang, Q., 2011. I-CASH: intelligently coupled array of SSDs and HDDs. Proc. 17th IEEE Int. Symp. on High Performance Computer Architecture, p.278-289.

[27]Rosenblum, M., Ousterhout, J.K., 1992. The design and implementation of a log-structured file system. ACM Trans. Comput. Syst., 10(1):26-52.

[28]Saxena, M., Swift, M.M., Zhang, Y., 2012. Flashtier: a lightweight, consistent and durable storage cache. Proc. European Conf. on Computer Systems, p.267-280.

[29]Schindler, J., Shete, S., Smith, K.A., 2011. Improving throughput for small disk requests with proximal I/O. Proc. 9th USENIX Conf. on File and Storage Technologies, p.133-147.

[30]Wachs, M., Abd-El-Malek, M., Thereska, E., et al., 2007. Argon: performance insulation for shared storage servers. Proc. 5th USENIX Conf. on File and Storage Technologies, p.61-76.

[31]Wu, G., He, X., 2012a. Reducing SSD read latency via NAND flash program and erase suspension. Proc. 10th USENIX Conf. on File and Storage Technologies, p.117-123.

[32]Wu, G., He, X., 2012b. Delta-FTL: improving SSD lifetime via exploiting content locality. Proc. 7th European Conf. on Computer Systems, p.253-266.

[33]Wu, G., He, X., Xie, N., et al., 2010. DiffECC: improving SSD read performance using differentiated error correction coding schemes. Proc. 18th Annual Meeting of the IEEE/ACM Int. Symp. on Modeling, Analysis and Simulation of Computer and Telecommunications Systems, p.57-66.

[34]Xu, Y., Jiang, S., 2011. A scheduling framework that makes any disk schedulers non-work-conserving solely based on request characteristics. Proc. 9th USENIX Conf. on File and Storage Technologies, p.119-132.