A new vertical fragmentation algorithm based on ant collective behavior in distributed database systems |
| |
Authors: | Mehdi Goli Seyed Mohammad Taghi Rouhani Rankoohi |
| |
Affiliation: | 1. Electrical & Computer Engineering Department, Shahid Beheshti University, Tehran, Iran
|
| |
Abstract: | Considering the existing massive volumes of data processed nowadays and the distributed nature of many organizations, there
is no doubt how vital the need is for distributed database systems. In such systems, the response time to a transaction or
a query is highly affected by the distribution design of the database system, particularly its methods for fragmentation,
replication, and allocation data. According to the relevant literature, from the two approaches to fragmentation, namely horizontal
and vertical fragmentation, the latter requires the use of heuristic methods due to it being NP-Hard. Currently, there are
a number of different methods of providing vertical fragmentation, which normally introduce a relatively high computational
complexity or do not yield optimal results, particularly for large-scale problems. In this paper, because of their distributed
and scalable nature, we apply swarm intelligence algorithms to present an algorithm for finding a solution to vertical fragmentation
problem, which is optimal in most cases. In our proposed algorithm, the relations are tried to be fragmented in such a way
so as not only to make transaction processing at each site as much localized as possible, but also to reduce the costs of
operations. Moreover, we report on the experimental results of comparing our algorithm with several other similar algorithms
to show that ours outperforms the other algorithms and is able to generate a better solution in terms of the optimality of
results and computational complexity. |
| |
Keywords: | |
本文献已被 SpringerLink 等数据库收录! |
|