A data accessing model for dynamic clustering of object-oriented databases
For dynamic clustering of objects, a data accessing model of online object-oriented databases is proposed in this paper. Dynamic clustering for object-oriented databases is an important issue in database management systems. Most of the previous researches have focused on the schemes of the dynamic clustering to optimal disk rearrangement, and not many works have been done on the behavior of the data requested pattern. For the data accessing sequence to consolidate and advance the state of research in object clustering problem, a statistical model of data accessing behavior is proposed. The behavior of data accessing sequence can be described by Poisson Processing Model (PPM). The criterion we suggested for the dynamic clustering strategy is an iterative numerical method, two preselected values need to be chosen, which are used to bind the deviations of the object-probabilities and the percentage of the objects in the object space that we can do without reclustering of the objects.
Author's Name: Chueh, H., Lin, N.P.
Volume: Volume 6
Issues: Issue 9
Keywords: Data accessing model, Disk rearrangement, Dynamic clustering, Object-oriented database, Poisson process model