spoke with Wilson Rivera, director of The Parallel and Distributed Computing Laboratory at the University of Puerto Rico, Mayaguez, about his lab's Grid testbed, which is being used to research and improve various areas of Grid computing. Rivera will be discussing this project at Gelato ICE, which takes place April 23-26 in San Jose, Calif.
Tell me a little about the PDCLab and about your history in Grid/distributed/high-performance computing.WILSON RIVERA:
The PDCLab was established in 2001 as part of the Program for Research in Computing and Information Science and Engineering (PRECISE), a project funded by the National Science Foundation to strengthen computing research at the University of Puerto Rico. Initially, the lab focused on the design and development of parallel algorithms targeting computational fluid dynamics and evolutionary computation. Since 2005, the PDCLab has focused its research activities almost exclusively to Grid computing technologies addressing fundamental research problems in Grid computing including automated grid deployment and adaptive Grid services.Gt:
On what kind of Grid and/or distributed computing projects is the PDCLab currently working?RIVERA:
Currently, we have several Grid computing related projects underway which include:
- WALSAIP -- The Wide Area Large Scale Automated Information Processing project aims at developing an infrastructure for the treatment of signal-based information arriving from physical sensors in a wide-area, large scale environment. It formulates a new conceptual model for treatment of signal information which accentuates a distributed space-time processing format. It permeates all other system substructures such as distributed sensor networks for signal acquisition, distributed databases for database management and distributed computing.
- STB-CASA -- This project, sponsored by the NSF Engineering Research Center for Collaborative Adaptive Sensing of the Atmosphere (CASA), focuses on integrating radar networks and Grid technologies to improve our ability to monitor the earth's lower atmosphere.
- ASPO -- The Adaptive Service Provision and Orchestration project addresses the problem of how multiple services should be orchestrated in a Grid environment to provide adaptive functionalities. The need for adaptation in Grid infrastructures arises due to both resource and service demand uncertainty. Next generation of Grid middleware must provide mechanisms to efficiently deal with uncertainty. Several key issues in this problem space are addressed to evolve, scale and respond to unpredictable service demands and events.
At the upcoming Gelato conference, you'll be speaking about your Grid testbed. Tell me a little about that project.RIVERA:
The PDCLab Grid testbed is not a production grid. It is indeed an experimental deployment of Grid computing technologies. The PDCLab Grid testbed has been thought to provide an easy-to-use infrastructure with flexibility to plug in new resources and testing of services. To achieve this goal, we have deployed a number of tools to facilitate administrative and end-user utilization via a package of scripts for configuration and installation and grid portals to access resources and services. To complement the spectrum of work in Grid computing technologies, we have developed specific research ideas including adaptive scheduling and data replication mechanisms. The ultimate goal is to apply these ideas in our Grid infrastructure and extrapolate them to other Grid-based infrastructures.Gt:
What kind of results have you seen thus far from the research you're conducting?RIVERA:
We have developed an adaptive scheduling algorithm, referred to as QB-MUF algorithm, to provide quality of service for wide-area, large-scale applications. The scheduling strategy focuses on providing high priority to jobs with low probability of failure based on Quality of Service (QoS) criteria. We are currently working on the deployment of this scheduling strategy as a Grid service on top of Globus toolkit 4.0.1.
We have also developed an information dispersal replication strategy to perform distributed management of data acquired by sensor networks. The proposed information dispersal algorithm shows a better access reliability than the traditional replication algorithms. It is also being deployed as a Grid service.Gt:
Can you discuss the hardware and software aspects of the Grid testbed? What types of middleware, processors, operating systems, etc., comprise this grid?RIVERA:
The PDClab Grid testbed aggregates a number of heterogeneous resources, including a Linux Beowulf Cluster that consists of 65 2-Way SMP Intel Pentium III, eight IA-64 Itanium servers, two IA-32 Pentium IV servers, one IA-32 Pentium III server and two Intel Xeon servers. The storage capacity is around of 4TB. It is being connected to a network of radars and sensors distributed around the island of Puerto Rico. The PDClab Grid testbed components run CentOS 4.2 (Linux) and the Globus Toolkit 4.0.1. Software associate to the pre-installation of Globus includes: OpenPBS, Torque, PosgreSQL, Apache Ant version 1.6.5, Java SDK version 1.5 and Jakarta Tomcat version 5.5.9.Gt:
Although grids are -- in theory, at least -- hardware and operating system agnostic, your institution is also a member of the Gelato Federation, which promotes Linux on Itanium. Why is it important for you? What makes Linux on Itanium an effective platform for high-performance computing?RIVERA:
We have demonstrated the benefits of the Itanium architecture on high- demand applications such as hyper-spectral imaging analysis. The heterogeneous nature of resources in our Grid testbed is an important issue since it posses a number of administrative and performance considerations. For example, configuration and deployment are quite different for Itanium-based resources versus I-32 based resources. In terms of execution of applications, it is difficult to hold transparency when submitting jobs to the grid. Applications targeting I-64 Itanium-based resources often require extra tuning efforts to achieve performance. As a consequence, an important effort in our research plan has been the development of tools to facilitate transparent access to these heterogeneous architectures.Gt:
How do you see Gelato members in Latin America contributing to science and technology development in the region?RIVERA:
Gelato Federation has been quite successful bringing together the top research institutions in Latin America. Institutions such as UFCG (Brazil), ITEM (Mexico), UC (Chile) and UB (Argentina) have a probed tradition in research contributions. I invite you to visit Gelato's Web site (www.gelato.org
) to learn more about the contributions of these institutions in varied areas such as high performance computing, digital libraries and scientific computing.Gt:
How does Grid computing play into this contribution?RIVERA:
Examples of large-scale Grid-related projects in Latin America include CLARA (Latin America Cooperation of Advanced Networks), supported by the European Commission through the ALICE project (Latin America Interconnected with Europe); EELA (E-infrastructure Shared between Europe and Latin America); and, recently, LAGrid (Latin American Grid).
In my opinion, there has been a lack of vision of the Latin American governments, including Puerto Rico, to fully support Grid computing adoption, contrary to government agencies in Europe and Asia. We cannot underestimate the impact that Grid computing technologies will have on science and technology development in the region and, ultimately, on our society.Gt:
Overall, how would you rate the success of your Grid testbed, and any other Grid projects at UPRM, and what does the future hold for Grid research at the university?RIVERA:
We did not want to be "yet another grid project" to provide computing facilities to end-users, so we have concentrated our efforts understanding Grid technologies and working on fundamental problems in Grid computing. I understand we are now in a good position of leadership for Grid initiatives at UPRM. That is my initial measure of success. However, there is too much work to do. I foresee exiting Grid projects at UPRM interconnecting intra- and inter-campus facilities. I expect a strong collaboration with our industrial partners on campus, such as IBM with LAGrid and Hewlett-Packard with the Utility Data Centers. I believe this is a unique opportunity to advance the industry of knowledge in Puerto Rico.Gt:
Is there anything else you'd like to add?RIVERA:
I am very thankful to have had the opportunity to talk about our Grid efforts at the PDCLab. I will keep GRIDtoday posted about our Grid initiatives at UPRM and, of course, I invite my colleagues around the world to contact us for potential collaborations. Our Web site is http://pdc.ece.uprm.edu
.About Wilson Rivera
Dr. Wilson Rivera obtained his Ph.D.in computational engineering from Mississippi State University, while working at the NSF Engineering Research Center for Computational Field Simulation. There, he concentrated on developing domain decomposition algorithms for solving time dependent partial differential equations with applications in Computational Fluid Dynamics. Rivera is an associate professor at the University of Puerto Rico, Mayaguez, (UPRM). He leads the Parallel and Distributed Computing Laboratory (PDCLab) at UPRM. His current funded projects address fundamental research problems in the areas of Grid computing (automated grid deployment, adaptive Grid services, dynamic resource management and grid performance) and workflow management (workflow modeling, metadata description and dynamic scheduling). Rivera is also the executive director for the Institute for Computing and Informatics Studies at UPRM and is a faculty member of the NSF Center for Subsurface Sensing and Imaging Systems (CenSSIS) and the NSF Center for Collaborative Adaptive Sensing of the Atmosphere (CASA).