A Low Cost Distributed Computing Approach to Pulsar Searches at a Small College

• Main Authors: Cantino, Andrew, Crawford, Fronefield, Dhital, Saurav, Dougherty, John P, Sherman, Reid
• Format: Journal Article
• Language: English
• Published: 07.07.2004
• Subjects: Computer Science - Distributed, Parallel, and Cluster Computing
• DOI: 10.48550/arxiv.cs/0407017

We describe a distributed processing cluster of inexpensive Linux machines developed jointly by the Astronomy and Computer Science departments at Haverford College which has been successfully used to search a large volume of data from a recent radio pulsar survey. Analysis of radio pulsar surveys requires significant computational resources to handle the demanding data storage and processing needs. One goal of this project was to explore issues encountered when processing a large amount of pulsar survey data with limited computational resources. This cluster, which was developed and activated in only a few weeks by supervised undergraduate summer research students, used existing decommissioned computers, the campus network, and a script-based, client-oriented, self-scheduled data distribution approach to process the data. This setup provided simplicity, efficiency, and "on-the-fly" scalability at low cost. The entire 570 GB data set from the pulsar survey was processed at Haverford over the course of a ten-week summer period using this cluster. We conclude that this cluster can serve as a useful computational model in cases where data processing must be carried out on a limited budget. We have also constructed a DVD archive of the raw survey data in order to investigate the feasibility of using DVD as an inexpensive and easily accessible raw data storage format for pulsar surveys. DVD-based storage has not been widely explored in the pulsar community, but it has several advantages. The DVD archive we have constructed is reliable, portable, inexpensive, and can be easily read by any standard modern machine.