Volume 7 Issue 1 - May 2016

  • 1. An efficient heuristic algorithm for reliable non distributed join processing in distributed databases

    Authors : Vivek Kumar, Vinod Kumar

    Pages : 270-278

    DOI : http://dx.doi.org/10.21172/1.71.039

    Keywords : Distributed Databases, Query processing, Heuristic Algorithms, Reliable Task Allocation.

    Abstract :

    In computing systems, the quality, in terms of assurance with some degree in its operations, is desired. This aspect of quality plays a greater role in distributed computing systems where multiple elements (processors, links etc.), with varied degree of capabilities, are involved. In the case of Distributed Database Systems, many measures at different levels are included. Keeping in view of assurance, the data units are replicated and transaction management protocols are designed. Reliability factor has also been included in deciding the strategy for the allocation of operations for performing the required query. This paper explores the reliability aspect when it is achieved through the use of redundant processors and communication links among the sites of the distributed database system and allocation of operation of distributed query with mission reliability as the objective. We have suggested a heuristic algorithm in contrast to [1] which has increased the computational efficiency relying on the fact that the addition and multiplication operations are more efficient than the exponential function calls and the product operation on the results of these calls. Programs are designed to implement both the original and the enhanced algorithms. Several experiments (simulated) are conducted and the results show a very considerable enhancement obtained by applying the present heuristic along with the concept of more efficient operations. A total of six different sets of queries have been considered for simulation. Each set has been explored for six different ranges of execution cost over 200 randomly generated sets, thus, resulting into a total of 7200 simulations. The total gain achieved in terms of number of nodes of the search tree is 84.22%. The execution time difference could not be clearly notioned below 5 joins. However, the approximated total gain in this regard is obtained as 53.84%.

    Citing this Journal Article :

    Vivek Kumar, Vinod Kumar, "An efficient heuristic algorithm for reliable non distributed join processing in distributed databases", Volume 7 Issue 1 - May 2016, 270-278