GRT: Global R-Trees
Year: 2013
Type: article
Abstract: As hardware technology advances, distributed architectures are becoming more and more complex.Writing software that fully and efficiently utilizes massively parallel systems can be an extremely difficult task.We seek to provide a shared memory view of distributed address space to ease the difficulty of implementing parallel algorithms.This thesis describes our parallel implementation of the R-Tree structure with this design goal in mind.Our approach differs from other parallel R-Tree implementations in that we map the linked structure of an R-Tree onto an array structure.We then leverage the Global Arrays framework to distribute the array structure across our parallel system.Global Arrays is a framework produced by Pacific Northwest National Laboratories that provides a shared memory view of multi-dimensional, distributed arrays.In order to improve the efficiency and performance of the Global Arrays framework, we have also implemented a caching system as a layer between our application and framework.Our cache provides automatic data replication across a distributed system as well as rudimentary support for write operations without affecting read performance.
Author: Alec Wiseman
Institution: The Ohio State University
Language: English
Subfield: Hardware and Architecture
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2619024160
