Summer 2007 REU students (Xabriel J. Collazo-Mojica, Javier Figueroa, Sean Leslie, Michael McFail, Barbara Morales, Joseph Naps, Alex Orta, Alain Esteva Ramirez, Ricardo Tirado, Ellen Wagner), graduate students and faculty mentors
Research and Education:
Participants: Joe Naps, Ellen Wagner, Jorge Guerra (Ph.D. Student), Luis Useche (Ph.D. Student), and Dr. Raju Rangaswami (mentor)
Project Description: Redundant Array of Inexpensive Disks (RAID) is the standard for large scale, fault-tolerant storage solutions. The various RAID levels combine space and/or performance optimization to meet the needs of the user. In the project, we developed UPGRAID, which can be seen as a new RAID level built upon RAID5. The idea behind UPGRAID is that, by allocating a small amount of disk space for RAID stripe replication, substantial gains can be seen in both reducing response time during normal operation and reducing overall reconstruction time when a drive fails. This decrease is important as reconstruction is the most vulnerable time for a RAID5 system because another disk failure would render the array useless.
Publications: Joseph Naps, Ellen Wagner, Jorge Guerra, Luis Useche, and Raju Rangaswami, "UPGRAID:Usage-based striPe replicatinG RAID", Under preparation
Participants: Ellen Wagner, Jorge Guerra (Ph.D. Student), Luis Useche (Ph.D. Student), Medha Bhadkamkar (Ph.D. Student), and Dr. Raju Rangaswami (mentor)
Project Description: Guaranteeing that systems perform correctly is imperative for system administrators and developers. Therefore there is a need for tools which test all parts and combinations of workloads. Systems handle a series of I/O requests for every action performed by the system, thus in order to assure consistency those I/O requests need to be handled properly. I/O requests can be either a read or a write and it is important that these operations get sent to the correct disk location and do not interfere with or corrupt existing data. Another important task for the tools is to simulate the environment in which the system will be run. System administrators will want to be prepared for the reaction of the system when run at any level of production. This second tool will give feedback as to what it predicts will happen when the user specifies a workload, and then both the developer and system administrator will know how the system will respond to various situations. This project involved building a storage block-layer 'integrity checker' and 'workload generator' test tools that can be used to test new block-layer extensions in the Linux storage stack.
Publications: Jorge Guerra, Luis Useche, Medha Bhadkamkar, Ellen Wagner*, and Raju Rangaswami, "Active Block Layer Extensions", Under preparation
Participants: Barbara Morales, Alain Esteva Ramirez, Tariq King (PhD student), and Dr. Peter J. Clarke (mentor)
Project Description: This project built on the work done by the REU 2006 students which involved investigating a component-based mechanism for safe dynamic adaptation that can be used in the validation of autonomic computing systems. The results of the project included: (1) an object-oriented (OO) design for the data structure which can be used to store the component dependencies, and build a safe adaptation graph that would be used to compute the sequence of adaptation steps; (2) a prototype of an autonomic system was developed including components such as orchestrating autonomic managers, touchpoint autonomic managers, and knowledge sources for a job scheduling application which incorporated self-configuration and self-optimization features.
Participants: Sean Leslie, Ricardo Tirado, Andrew Allen (PhD student) and Dr. Peter J. Clarke (mentor)
Project Description: The project investigated how to extend the functionality of a Network Communication Broker (NCB), a layer in the Communication Virtual Machine (CVM) that interfaces with the underlying networks, to utilize multiple communication APIs and the dynamic configuration of these APIs in NCB. An extensive review of the following communication APIs was performed: NCB, Skype, Libjingle, Google Talk's C++, Eclipse Communication Framework and Java MSN Messenger Library (JML). The results of the project included: (1) dynamic models (UML activity diagrams) and static models (UML class diagrams) for the self-configuring capability of the NCB. A prototype of the self-configuring NCB was also built.
Participants: Javier Figueroa, Xabriel J Collazo Mojica, Michael McFail, Alexander Orta, David Villegas (PhD Student), Javier Muñoz (PhD Student), Diego Lopez (PhD Student), and Dr. S. Masoud Sadjadi (mentor)
Project Description: The impact of hurricanes is so devastating throughout different levels of society that there is a pressing need to provide a range of users with accurate and timely information that can enable effective planning for and response to potential hurricane landfalls. The Weather Research and Forecasting (WRF) code is the latest numerical model that has been adopted by meteorological services worldwide. The current version of WRF has not been designed to scale out of a single organization's local computing resources. However, the high resource requirements of WRF for fine-resolution and ensemble forecasting demand a large number of computing nodes, which typically cannot be found within one organization. Therefore, there is a pressing need for the Grid-enablement of the WRF code such that it can utilize resources available in partner organizations. In this project, we investigate Grid enablement of WRF by leveraging our work in transparent shaping, GRID superscalar, profiling, code inspection, code modeling, meta-scheduling, and job flow management.