Areliable and energy-efficient storage systemwith erasure coding cache

Abstract

In modern energy-saving replication storage systems, a primarygroup of disks is always powered up to serve incoming requests whileother disks are often spun down to save energy during slack periods.However, since new writes cannot be immediately synchronized intoall disks, system reliability is degraded. In this paper, we developa high-reliability and energy-efficient replication storage system,named RERAID, based on RAID10. RERAID employs part of the free spacein the primary disk group and uses erasure coding to construct a codecache at the front end to absorb new writes. Since code cache supportsfailure recovery of two or more disks by using erasure coding, RERAIDguarantees a reliability comparable with that of the RAID10 storagesystem. In addition, we develop an algorithm, called erasure codingwrite (ECW), to buffer many small random writes into a few large writes,which are then written to the code cache in a parallel fashion sequentiallyto improve the write performance. Experimental results show that RERAIDsignificantly improves write performance and saves more energy thanexisting solutions.

Keywords

Reliability / Energy-efficient / Storage system / Erasure coding / Cache management

References

[1] Amazon, 2007. Amazon S3: Object Storage Built toStore and Retrieve Any Amount of Data from Anywhere.

[2] Amur , H., Cipar , J., Gupta , V., , 2010. Robust andflexible power-proportional storage. Proc.1st ACM Symp. on Cloud Computing, p.217–228.

[3] Bhadkamkar , M., Guerra , J., Useche , L., , 2009. BORG: Block-reORGanizationfor self-optimizing storage systems. Proc.Usenix Conf. on File and Storage Technologies, p.183–196.

[4] Blaum , M., Brady , J., Bruck , J., , 1994. EVENODD:an optimal scheme for tolerating double disk failures in RAID architectures. Proc. 21st Int. Symp. on Computer Architecture, p.245–254.

[5] Borthaku , D., 2010. What is Apache Hadoop?

[6] Chen , Y., Hsu , W., Young , H., 2000. Logging RAID—anapproach to fast, reliable, and low-cost disk arrays. Euro-Par, p.1302–1312.

[7] Colarelli , D., Grunwald , D., 2002. Massive arrays of idle disks for storage archives. Proc. ACM/IEEE Conf. on Supercomputing, p.1–11.

[8] Corbett , P., English , B., Goel , A., , 2004. Row-diagonal parity for double disk failure correction. Proc. 3rd USENIX Conf. on File and Storage Technologies, p.1–14.

[9] Department of Energy, 2012. NETL shares computing speed, efficiencyto tackle barriers. Fossil Energy Today, 1(6):1–3.

[10] EMC, 2008. ATMOS: Big. Smart. Elastic.

[11] Eom , H., Hollingsworth , J.K., 2000. Speed vs. accuracy in simulation for I/O-intensiveapplications. Proc. 14th Int. Paralleland Distributed Processing Symp., p.315–322.

[12] Ghemawat , S., Gobioff , H., Leung , S., 2003. The GoogleFile System. Proc. 19th ACM Symp. on OperatingSystems Principles, p.29–43.

[13] Hu , Y., Yang , Q., 1996. DCD—disk caching disk: a new approach for boostingI/O performance. ACM SIGARCH Comput. Archit. News, 24(2):169–178.

[14] Li , D., Wang , J., 2004. EERAID: energy-efficient redundant and inexpensive diskarray. Proc. 11th ACM SIGOPS European Workshop, p.29.

[15] Li , D., Wang , J., 2006. eRAID: a queuing model based energy saving policy. Proc. IEEE Int. Symp. on Modeling, Analysis, andSimulation of Computer and Telecommunication Systems, p.77–86.

[16] Lu , L., Varman , P.J., Wang , J., 2007. DiskGroup:energy efficient disk layout for RAID1 systems. Proc. Int. Conf. on Networking, Architecture, and Storage, p.233–242.

[17] Mao , B., Feng , D., Jiang , H., , 2008. GRAID: agreen RAID storage architecture with improved energy efficiency andreliability. Proc. IEEE Int. Symp. on Modeling,Analysis, and Simulation of Computer and Telecommunication Systems, p.113–120.

[18] Menon , J., 1995. A performanee comparison of RAID-5and log-struetured arrays. Proc. 4th IEEEIn. Symp. on High Performance Distributed Computing, p.167–178.

[19] Patterson , D.A., Gibson , G., Katz , R.H., 1988. A casefor redundant arrays of inexpensive disks (RAID). Proc. ACM SIGMOD Int. Conf. on Management of Data, p.109–116.

[20] Plank , J.S., Xu , L.H., 2006. Optimizing Cauchy Reed-Solomon codes for fault-tolerantstorage applications. Proc. 5th Int. Symp.on Network Computing and Applications, p.173–180.

[21] Pinheiro , E., Bianchini , R., 2004. Energy conservation techniques for disk array-based servers. Proc. 18th Annual Int. Conf. on Supercomputing, p.68–78.

[22] Pinheiro , E., Weber , W.D., Barroso , L.A., 2007. Failuretrends in a large disk drive population. Proc. 5th USENIX Conf. on File and Storage Technologies, p.17–28.

[23] Soundararajan , G., Prabhakaran , V., Balakrishnan , M., , 2010. Extending SSD lifetimes with disk-based write caches. Proc. 8th USENIX Conf. on File and Storage Technologies, p.101–114.

[24] Stodolsky , D., Gibson , G., Holland , M., 1993. Paritylogging overcoming the small write problem in redundant disk arrays. ACM SIGARCH Comput. Architect.News, 21(2):64–75.

[25] Thereska , E., Donnelly , A., Narayanan , D., 2011. Sierra:practical power-proportionality for data center storage. Proc. 6th Conf. on Computer Systems, p.169–182.

[26] Wang , J., Zhu , H.J., Li , D., 2008. eRAID:conserving energy in conventional disk-based RAID system. IEEE Trans. Comput., 57(3):359–374.

[27] Weil , S., Brandt , S.A., Miller , E.L., , 2006. Ceph: a scalable, high-performance distributed file system. Proc. 7th Conf. on Operating Systems Design andImplementation, p.307–320.

[28] Wilkes , J., Golding , R., Staelin , R., , 1996. The HP AutoRAID hierarchical storage system. ACM Trans. Comput. Syst., 14(1):108–136.

[29] Xin , Q., Miller , E.L., Schwarz , T., , 2003. Reliability mechanisms for very large storage systems. Proc. 20th IEEE/11th NASA Goddard Conf. on MassStorage Systems and Technologie, p.146–156.

[30] Xu , L.H., Bruck , J., 1999. X-code: MDS array codes with optimal encoding. IEEE Trans. Inform. Theory, 45(1):272–276.

[31] Yue , Y., Tian , L., Jiang , H., , 2010. RoLo: a rotatedlogging storage architecture for enterprise data centers. Proc. IEEE 30th Int. Conf. on Distributed ComputingSystems, p.293–304.

[32] Yue , Y., He , B., Tian , L., , 2016. Rotated loggingstorage architectures for data centers: models and optimizations. IEEE Trans. Comput., 65(1):203–215.

[33] Zhu , Q., Chen , Z., Tan , L., , 2005. Hibernator:helping disk arrays sleep through the winter. Proc. 20th ACM Symp. on Operating Systems Principles, p.177–190.

RIGHTS & PERMISSIONS

Zhejiang University and Springer-Verlag GmbHGermany