الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 16 |  |  |
| | | from | 16 | | |
Abstract
Cloud computing is now seen as a promising, cost-effective paradigm to support the execution of compute- and data-intensive spatial applications. The most existing spatial index structures for Cloud platform are based on R-tree, but there is not any study for evaluating R-Tree variants yet, especially PR-Tree. The goal of this thesis is the performance estimation of using cloud computing for indexing huge spatial datasets. In this work, we studied the performance of R-Tree and PR-Tree applications on various instances of Amazon EC2 cloud platform and also on the PC. The study of the performance has been done for both data structures by making the run time tests and the memory consumption tests.
We conclude from our experimental results that when using PR-Tree for spatial indexing the performance of the indexing and querying data is better than using R-Tree in case of using the same dataset on the PC . Also on the cloud the results also show that PR-Tree always gives better performance in the cloud than the PC and R-Tree. In some cases gives better performance in the PC than the cloud, but still better than R-Tree. The performance on the cloud in some cases is a little better because the application is not optimized to run on configurations with multiple CPU cores. In case of memory consumption tests, the results are the same on the PC and on the various cloud instance and PR-Tree consumes more memory than R-Tree.