Abstract:
Existing research on high-availability(HA) of data stream systems mainly focused on restoring single node failure effectively, replacing the failed nodes with redundant b...Show MoreMetadata
Abstract:
Existing research on high-availability(HA) of data stream systems mainly focused on restoring single node failure effectively, replacing the failed nodes with redundant backups to make the system go well. However, how to improve availability of DDSMS from the perspective of operator redundancy and multiple failures are paid less attention. In this paper, an improved HA mechanism based on operator redundancy in DDSMS is researched, a combined method for evaluating system overhead is proposed, and a way to choose the maximum redundancy is presented. The simulation verifies that this mechanism can provide a better synchronization service between the primary and the backup operator, and the chosen maximum redundancy can provide the maximum HA without affecting system performance.
Date of Conference: 10-15 November 2013
Date Added to IEEE Xplore: 27 March 2014
ISBN Information: