Project View

Optimised Data Storage Caching with High Availability Data Delivery within a Distributed Storage Network

Principal Investigator: Dr X. Peng, Aston University

Researchers: Dr X. Zhang, R. Haywood and T. Porter, Aston University

Sponsors: EPSRC &  Xyratex

 

Summary

 

The current technologies and network architectures for the storage network support high quality multimedia data delivery for mobile or remote computing users in the distributed or wireless environments but at a high cost of ownership. As more and more network subscribers join existing distributed storage network the costs of maintaining the reliability and data delivery performance will grow significantly. The project is focussed on a novel distributed storage architecture which addresses the concerns over scalability, reliability and speed. It aims to tackle these problems by introducing a combined strategy that allows the storage network to efficiently distribute data objects over the storage nodes through intelligent caching, eliminate the need for retransmission by properly controlling packet loss, and subsequently ensure reliable and speedy delivery of high-quality data to mobile computing users. In the new storage system proposed, storage nodes or multiple mirror sites store a sub-set of properly encoded packets of the files with a substantially large size.  Upon the request of a file at an access node, the neighbouring nodes will contribute their sub-set of the packets towards the reconstruction of the file requested. This scheme saves storage space, distributes traffic loads across different links, and enables reliable data transfer from source nodes to the clients in parallel. It is intended that the performance gains achieved by using the new scheme will be examined against the costs and overheads involved for different types and scales of applications. The research will be carried out at Aston University and partnered by Xyratex, a technology and market leader in the OEM storage network industry.

 

Objectives

 

*     Investigate the effect of packet loss on the delivery latency and system throughput for streaming media services in distributed networks.

*    Investigate the effect of wireless channels (both fixed and mobile) and user mobility on the availability of transmission links, in terms of the failure rate, and the network resource distribution and management.

*      Construct MDS erasure codes optimised for efficient large file dispersal schemes and for the provision of fault-tolerant data retrieval in distributed and wireless storage networks.

*     Devise efficient graph-colouring algorithms to optimise data distribution within a network. These will contrast the performance gains against the level of local storage requirements in the network.

*     Develop computationally efficient decoding techniques and implement the data processing interfaces in storage units using FPGA for encoding/decoding during the processes of content distribution and file reconstruction.

*     Integrate the communication and networking protocols required in support of deploying the proposed coding and distribution schemes into the existing network storage standards or the emerging OSD standard, and enable QoS management in the distributed storage network.

 

Publications

 

X. Zhang, and X.-H. Peng, A testbed of erasure coding on video streaming system over lossy networks, in Proc. IEEE 7th International Symposium on Communications and Information Technologies (ISCIT), Oct. 2007.

 

R. Haywood, and X.-H. Peng On packet loss performance under varying network conditions with path diversity, in Proc. ACM International Conference on Advanced Infocomm Technology (ICAIT), July 2008.

 

X. Zhang, X.-H. Peng, R. Haywood, and T. Porter, Robust video transmission over lossy network by exploiting H.264/AVC data partitioning, in Proc. IEEE 5th International Conference on Broadband Communications, Networks and Systems (BROADNETS), Sept. 2008.

 

R. Haywood, and X.-H. Peng, Effect of path diversity on the loss performance of UDP packets over the Internet, in Proc. IEEE 5th International Conference on Broadband Communications, Networks and Systems, Sept. 2008.

 

X. Zhang, X.-H. Peng, D. Wu, T. Porter, and R. Haywood, A hierarchical unequal packet loss protection scheme for robust H.264/AVC transmission, in Proc. IEEE 6th Annual Consumer Communications & Networking Conference (CCNC), Jan. 2009.

 

X. Zhang, and X.-H. Peng, An unequal packet loss protection scheme for H.264/AVC video transmission, in Proc. IEEE 23rd International Conference on Information Networking (ICOIN), Jan. 2009.

R. Haywood, S. Mukherjee, and X.-H. Peng, Investigation of H.264 video streaming over an IEEE 802.11e EDCA wireless testbed, in Proc. IEEE International Conference on Communications (ICC), June 2009.

X.-H. Peng and F.G. Farrell, On construction of the (24,12,8) Golay codes, IEEE Transactions on Information Theory, vol. 52,4 pp. 3669-3675, Aug. 2006.

X.-H. Peng, Erasure-control coding for distributed networks, IEE Proceedings on Communications, vol. 152, pp. 1075-1080, Dec. 2005.

R. Haywood, and X.-H. Peng, Optimal caching for partitioned video, in Proc. IEEE International Symposium on Multimedia (ISM), Dec. 2009.

T. Porter, and X.-H. Peng, Effective video content distribution by combining TCP with adaptive FEC coding, in Proc. IEEE International Symposium on Broadband Multimedia Systems and Broadcasting (BMSB), March 2010.

X.-H. Peng, Bounds on end-to-end performance of networks employing erasure control coding in Proc. IEEE International Symposium on Information Theory and Applications (ISITA), Oct. 2010

T. Porter, and X.-H. Peng, An objective approach to measuring video playback quality in lossy networks using TCP, IEEE Communications Letters, Vol. 15 Issue 1, Jan. 2011, pp. 76-78.