li {margin-left:25px;margin-bottom:10px}
Project Summary
This research project is an attempt to explore the challenges associated with using a cloud computing framework to launch distributed and parallelized GIS computations. Example GIS computations include terrain model construction and conversion, topographic characterization, and terrain visibility. I will be using existing algorithms and converting existing implementations to run on a cloud computing framework. The target cloud computing framework is provided by Amazon Web Services as described below in the Methodology section.
Anticipated Project Outcomes
- The production of a distributed, parallel GIS “engine” implemented using the AWS cloud computing framework – The engine will have parallelized implementations of common GIS computations. The front-end for the engine will be a set of web services that allow authenticated access to the parallel computation engine. For example, a web application could initiate viewshed analysis using the cloud computing framework by issuing the request:
http://ip_address_or_domain/gisop=viewshed&pt=34.312323,-86.323435&bbox=…
- An assessment of the performance and scalability of the engine using the AWS computing framework – I will perform a series of structured experiments to benchmark the performance of the GIS operations as the number of AWS compute nodes is increased.
- Increased understanding of the AWS cloud computing framework – As part of my work, I will gain experience and a better understanding of how to use the AWS cloud computing framework. This will enhance my ability to use the AWS cloud as part of courses that I teach as well as for future research and commercial projects.
- Appropriate research publications documenting the results of my research – I will look for an appropriate conference and/or journal to publish the outcome of my research project in addition to my faculty blog updates on the status of the project.
Motivation
As the quantity, quality, and diversity of GIS data continues to grow, the time required to perform common GIS operations is also increasing at an alarming rate. Consider the example shown in the two pictures below …
Elevation data accurate to 30 meters for the Southeastern United States
Over 700MB of elevation data
Elevation data cropped to the state boundary for Florida
Export run-time … nearly 5 days on an overclocked Core 2 Quad processor (3.7GHz)!
Methodology
I have divided this research project into three phases to be completed over the course of the next year as described below:
- Phase I – Learning the ropes with the AWS cloud computing framework.
- Phase II – Designing the engine for distributed, parallelized GIS computations.
- Phase III – Benchmarking the performance of the engine under various cloud framework configurations