Information Technology Reference
In-Depth Information
The overhead of SRDU with one redundant node is 11,123 blocks while the total overhead
for 10 redundant nodes is only 12,923 blocks. Therefore, the overhead for updating additional
redundant nodes can be substantially reduced by computing and sending partial results.
15.6 Summary
In this chapter, we investigated the system expansion problem in systems employing striped
storage, such as disk arrays, RAID, parallel servers, and even peer-to-peer systems. We pre-
sented efficient algorithms to carry out the two essential operations in expanding the system
to include more nodes, namely a Row-Permutated Data Reorganization (RPDR) algorithm
for reorganizing and redistributing media data blocks to the new nodes so that streaming load
balance can be restored; and a Sequential Redundant Data Update (SRDU) algorithm for the
efficient update of redundant data to support the new data organization after one or more new
nodes are added.
Note that this chapter did not address issues in the transmission of the data blocks, redundant
blocks, and partial results during the redundant data update process. As the system is online
serving active media streams, some of the nodes from time to time may not have sufficient
bandwidth to receive or transmit the data blocks. Thus proper scheduling must be done to avoid
congesting a receiver node while maximizing utilization of the idle bandwidth in the system
to shorten the system reorganization time.
References
[1] D.A. Patterson, G. Gibson, and R.H. Katz, A Case for Redundant Arrays of Inexpensive Disks (RAID), Proc. of
1988 ACM SIGMOD Conference on Management of Data , Chicago, June 1988.
[2] S. Ghandeharizadeh and D. Kim, On-line Reorganization of Data in Scalable Continuous Media Servers, Pro-
ceedings 7th International Conference on Database and Expert Systems Applications , Sept. 1996.
[3] A. Goel, C. Shahabi, S.-Y. Yao, and R. Zimmerman, SCADDAR: An Efficient Randomized Technique to Reor-
ganize Continuous Media Blocks, Proc. International Conference on Data Engineering , 2002.
[4] J.S. Plank, A Tutorial on Reed-Solomon Coding for Fault-Tolerance in RAID-like Systems, Software Practice
and Experience , vol. 27, no. 9, Sept. 1997, pp. 995-1012.
[5] L. Rizzo, Effective Erasure Codes for Reliable Computer Communication Protocols, ACM Computer Communi-
cation Review , vol. 27, no. 2, Apr. 1997, pp. 24-36.
[6] A.J. McAuley, Reliable Broadband Communication Using a Burst Erasure Correcting Code, Proceedings of the
ACM Symposium on Communications Architectures and Protocols , Sept. 1990, pp. 297-306.
Search WWH ::




Custom Search