A5069008173- Distributed and Parallel Computing Systems
- Scientific Computing and Data Management
- Parallel Computing and Optimization Techniques
- Advanced Data Storage Technologies
- Software System Performance and Reliability
- Cloud Computing and Resource Management
- Distributed systems and fault tolerance
- Peer-to-Peer Network Technologies
- Research Data Management Practices
- Network Security and Intrusion Detection
- Data Quality and Management
- Genetics, Bioinformatics, and Biomedical Research
- Semantic Web and Ontologies
- Advanced Database Systems and Queries
- Interconnection Networks and Systems
- Software-Defined Networks and 5G
- Anomaly Detection Techniques and Applications
- Big Data and Business Intelligence
- Service-Oriented Architecture and Web Services
- Network Traffic and Congestion Control
- Advanced Data Compression Techniques
- Plant Disease Management Techniques
- Green IT and Sustainability
- Internet Traffic Analysis and Secure E-voting
- Real-Time Systems Scheduling
The University of Texas at Austin
2023
Indiana University
2017-2022
Indiana University Bloomington
2016-2021
Indiana State University
2021
Purdue University West Lafayette
2019
Woods Hole Oceanographic Institution
2009-2012
Energy Sciences Network
2010
Argonne National Laboratory
2001-2007
University of Chicago
2006-2007
University of Edinburgh
2006-2007
Ensembles of distributed, heterogeneous resources, also known as computational grids, have emerged critical platforms for high-performance and resource-intensive applications. Such provide the potential applications to aggregate enormous bandwidth, power, memory, secondary storage, other resources during a single execution. However, achieving this performance in dynamic, environments is challenging. Recent experience with distributed indicates that adaptivity fundamental application dynamic...
Heterogeneous networks are increasingly being used as platforms for resource-intensive distributed parallel applications. A critical contributor to the performance of such applications is scheduling constituent application tasks on network. Since often resources cannot be brought under control a single global scheduler, must scheduled by user. To obtain best performance, user take into account both application-specific and dynamic system information in developing schedule which meets his or...
Monitoring and information services form a key component of distributed system, or Grid. A quantitative study such can aid in understanding the performance limitations, advise deployment monitoring help evaluate future development work. To this end, we three for systems: Globus Toolkit/spl reg/ Discovery Service (MDS2), European Data Grid Relational Architecture (R-GMA) Hawkeye, part Condor project. We perform experiments to test their scalability with respect number users, resources amount...
In heterogeneous and dynamic environments, efficient execution of parallel computations can require mappings tasks to processors whose performance is both irregular (because heterogeneity) time-varying dynamicity). While adaptive domain decomposition techniques have been used address resource capabilities, temporal variations in those capabilities seldom considered. We propose a conservative scheduling policy that uses information about expected future variance produce more data mapping...
As Data Grids become more commonplace, large data sets are being replicated and distributed to multiple sites, leading the problem of determining which replica can be accessed most efficiently. The answer this question depend on many factors, including physical characteristics resources load behavior CPUs, networks, storage devices that part end-to-end path linking possible sources sinks. We develop a predictive framework combines (1) integrated instrumentation collects information about...
The dynamic nature of a resource-sharing environment means that applications must be able to adapt their behavior in response changes system status. Predictions future performance can used guide such adaptations. In this paper, we present and evaluate several new one-step-ahead low-overhead time series prediction strategies track recent trends by giving more weight data. We results show tendency model with different ascending descending performs best among all studied. A comparative study...
The recent proliferation of Data Grids and the increasingly common practice using resources as distributed data stores provide a convenient environment for communities researchers to share, replicate, manage access copies large datasets. This has led question which replica can be accessed most efficiently. In such environments, fetching from one several locations requires accurate predictions end-to-end transfer times. answer this depend on many factors, including physical characteristics...
Accurate performance predictions are difficult to achieve for parallel applications executing on production distributed systems. Conventional point-valued parameters and prediction models often inaccurate since they can only represent one point in a range of possible behaviors. The authors address this problem by allowing characteristic application system data be represented set values their probabilities, which call stochastic values. They give practical methodology using as adaptable...
Monitoring and information services form a key component of distributed system, or grid. A quantitative study such can aid in understanding the performance limitations, advise deployment monitoring help evaluate future development work. To this end, we examined Globus Toolkit/spl reg/ Discovery Service (MDS2) by instrumenting its main using NetLogger. Our shows strong advantage to caching prefetching data, as well need have primary components at well-connected sites.
The Globus Toolkit Monitoring and Discovery System (MDS4) defines implements mechanisms for service resource discovery monitoring in distributed environments. MDS4 is distinguished from previous similar systems by its extensive use of interfaces behaviors defined the WS-Resource Framework WS-Notification specifications, deep integration into essentially every component Toolkit. We describe architecture Web that allow users to discover resources services, monitor states, receive updates on...
The increasingly common practice of replicating datasets and using resources as distributed data stores in grid environments has led to the problem determining which replica can be accessed most efficiently. Due diverse performance characteristics load variations several components end-to-end path linking these various locations, selecting a from among many requires accurate prediction information transfer times between sources sinks. In this paper we present system that is based on...
Article Free Access Share on Stochastic scheduling Authors: Jennifer M. Schopf Computer Science Department, Northwestern University UniversityView Profile , Francine Berman Dept. of and Engineering, California, San Diego DiegoView Authors Info & Claims SC '99: Proceedings the 1999 ACM/IEEE conference SupercomputingJanuary Pages 48–eshttps://doi.org/10.1145/331532.331580Published:01 January 1999Publication History 32citation747DownloadsMetricsTotal Citations32Total Downloads747Last 12...
The design and implementation of a national computing system data grid has become reachable goal from both the computer science computational point view. A distributed infrastructure capable sophisticated functions can bring many benefits to scientific work, but poses challenges, technical socio‐political. Technical challenges include having basic software tools, higher‐level services, functioning pervasive security, standards, while socio‐political issues building user community, adding...
Today's system monitoring tools are capable of detecting failures such as host failures, OS errors, and network partitions in near-real time. Unfortunately, the same cannot yet be said end-to-end distributed software stack. Any given action, for example, reliably transferring a directory files, can involve wide range complex interrelated actions across multiple pieces software: checking user certificates permissions, getting details all performing third-party transfers, understanding re-try...
Virtual organizations (VOs), communities that enable coordinated resource sharing among multiple sites, are becoming more prevalent in the high-performance computing community. In order to promote cross-site usability, most VOs prepare service agreements include a minimum set of common functionality, starting with software stack and evolving into complicated interoperability agreements. VO often difficult verify maintain, however, because sites dynamic autonomous. Automated verification is...
Identifying and diagnosing anomalies in application behavior is critical to delivering reliable application-level performance. In this paper we introduce a strategy detect diagnose the possible reasons behind them. Our approach extends traditional window-based by using signal-processing techniques filter out recurring, background fluctuations resource behavior. addition, have developed diagnosis technique that uses standard monitoring data determine which related changes may cause anomalies....