Publication at Faculty of Mathematics and Physics |
2011
Abstract
One possible way of achieving the parallelization in the data processing is the use of data streams. Data in such systems are considered to be a flow of reasonably small data packets, so that different packets can be processed by different computational units in parallel.
Unfortunately, some operations are not suitable for use in data stream processing, because they require the whole data to be accessible at a time, and sorting is one of them. These operations reduce the level of parallelism and scalability as well, which could be potentially achieved.
In this paper we propose implementation of sorting algorithm which is suitable for use in data stream processing.