iVISA: a framework for flexible layout block-level storage system

High performance, high bandwidth, and scalable network storage system is the requirement tendency of most data-intensive applications. Distributed RAID, as a storage architecture, is widely used in cluster and distributed computing environments. But the data placement is mechanical and penalty for maintaining consistency is suffered. In this paper we propose a novel block-level storage architecture called iVISA (iSCSI based virtual interface storage architecture) which employs the concepts both distributed RAID and virtualization storage. All storage resources distribute in a high performance VI network and are managed by a metadata server. A number of iSCSI target nodes act as the entrances of iVISA system for accessing through iSCSI connections. The storage resources are dynamically allocated to users during I/O accessing. The users' logical block addresses are intelligently and flexibly mapped to the physical block addresses based on the current workload of storage nodes and data layout. The evaluation result shows that better performance can be achieved by allocation and mapping dynamically and flexibly with optimized allocation strategy. A hybrid strategy considering both data layout and load of storage nodes has a 30% higher I/O performance than conventional distributed RAID.

[1]  Craig A. N. Soules,et al.  A Two-Tiered Software Architecture for Automated Tuning of Disk Layouts (CMU-CS-03-130) , 2003 .

[2]  Hai Jin,et al.  Orthogonal Striping and Mirroring in Distributed RAID for I/O-Centric Cluster Computing , 2002, IEEE Trans. Parallel Distributed Syst..

[3]  Greg Lehey The Vinum Volume Manager , 1999, USENIX Annual Technical Conference, FREENIX Track.

[4]  Giovanni Chiola,et al.  Using a Gigabit Ethernet cluster as a distributed disk array with multiple fault tolerance , 2003, 28th Annual IEEE International Conference on Local Computer Networks, 2003. LCN '03. Proceedings..

[5]  Angelos Bilas,et al.  Clotho: Transparent Data Versioning at the Block I/O Level , 2004, MSST.

[6]  Angelos Bilas,et al.  Violin: a framework for extensible block-level storage , 2005, 22nd IEEE / 13th NASA Goddard Conference on Mass Storage Systems and Technologies (MSST'05).

[7]  Mario Lauria,et al.  RAAC: an architecture for scalable, reliable storage in clusters , 2004, 2004 IEEE International Conference on Cluster Computing (IEEE Cat. No.04EX935).

[8]  Kenneth Salem,et al.  Adaptive block rearrangement , 1993, TOCS.

[9]  Gregory R. Ganger,et al.  Object-based storage , 2003, IEEE Commun. Mag..

[10]  Hai Jin,et al.  Distributed RAID A New Multiple Copy Algorithm , 2002 .

[11]  Angelos Bilas,et al.  Violin: A Framework for Extensible Block-Level Storage , 2005, MSST.

[12]  Yanping Zhao,et al.  HyLog: A High Performance Approach to Managing Disk Layout , 2004, FAST.

[13]  Friedhelm Meyer auf der Heide,et al.  V: Drive - Costs and Benefits of an Out-of-Band Storage Virtualization System , 2004, MSST.