Person:
Stasko, John T.

Permanent Link
https://hdl.handle.net/1853/71735
Associated Organization(s)
Organizational Unit
ORCID
ArchiveSpace Name Record

Filters

1994 - 1997 5
4 1
5
5
5
Reset filters

Settings
Author: Topol, Brad Byer × Type: Text ×

Publication Search Results

Now showing 1 - 5 of 5
Thumbnail Image
Item

Robust State Sharing for Wide Area Distributed Applications

1997 , Topol, Brad Byer , Ahamad, Mustaque , Stasko, John T.

In this article, we present the Mocha wide area computing infrastructure we are currently developing. Mocha provides support for robust shared objects on heterogeneous platforms, and utilizes advanced distributed shared memory techniques for maintaining consistency of shared objects that are replicated at multiple nodes to improve performance. In addition, our system handles failures that we feel will be common in wide area environments. For example, to ensure that the state of an object is not lost due to a node failure, updated state of the object can be disseminated to several other nodes. The overhead of such state dissemination can be controlled based on the level of availability needed for shared objects. We have used an approach that makes use of multiple communication protocols to improve the efficiency of shared object state transfers in Mocha. We also provide an empirical evaluation of our prototype implementation for both local and wide area networks and present a sample home service application written with the system.

View
Thumbnail Image
Item

Integrating Visualization Support Into Distributed Systems

1994 , Topol, Brad Byer , Stasko, John T. , Sunderam, Vaidy

Visualization and animation tools may become extremely important aids in the understanding, verification, and performance tuning of parallel computations. Presently, however, the use of visualization has had only a limited use for enhancing parallel computation. We hypothesize that one of the primary reasons for the limited use of visualization tools in parallel program development is the difficulty of acquiring the information necessary to drive the visual display. Our approach to this impediment focuses on integrating visualization support directly into a distributed computing system. Central to this integration is the addition of a logical clock that prevents the timestamps of events from violating causality. The implementation requires the ``piggybacking'' of a negligible amount of extra header information on system messages and the impact on performance is minimal. This results in a system that produces useful visualizations with no extra effort required by the applications programmer. Also integrated into the distributed system is support which simplifies the creation of programmer-defined, application-specific visualizations, unique to each new parallel program developed.

View
Thumbnail Image
Item

Monitoring and Visualization in Cluster Environments

1996 , Topol, Brad Byer , Stasko, John T. , Sunderam, Vaidy

Cluster computing has evolved into a popular and effective mode of high performance computing. Cluster environments are intrinsically different from hardware multiprocessors, and hence require a different approach to measuring and characterizing performance, monitoring an application's progress, and understanding program behavior. In this article, we present the design and implementation of PVaniM, an experimental visualization environment we have developed for the PVM network computing system. PVaniM supports a two-phase approach whereby on-line visualization focuses on large-grained events that are influenced by and relate to the dynamic cluster environment, and postmortem visualization provides for detailed program analysis and tuning. PVaniM's capabilities are illustrated via its use on several applications and it is compared with other visualization environments developed for cluster computing. Our experiences indicate that for several classes of applications, the two-phase visualization scheme can provide more insight into the behavior, efficiency, and operation of distributed and parallel programs in cluster environments.

View
Thumbnail Image
Item

Integrating Visualization Support Into Distributed Computing Systems

1994 , Topol, Brad Byer , Stasko, John T. , Sunderam, Vaidy

Visualization and animation tools may become extremely important aids in the understanding, verification, and performance tuning of parallel computations. Presently, however, the use of visualization has had only a limited use for enhancing parallel computation. We hypothesize that one of the primary reasons for the limited use of visualization tools in parallel program development is the difficulty of acquiring the information necessary to drive the visual display. Our approach to this impediment focuses on integrating visualization support directly into a distributed computing system. Central to this integration is the addition of a logical clock that prevents the timestamps of events from violating causality. The implementation requires the "piggybacking" of a negligible amount of extra header information on system messages and the impact on performance is minimal. This results in a system that produces useful visualizations with no extra effort required by the applications programmer. Also integrated into the distributed system is support which simplifies the creation of programmer-defined, application-specific visualizations, unique to each new parallel program developed.

View
Thumbnail Image
Item

The Dual Timestamping Methodology for Visualizing Distributed Applications

1995 , Topol, Brad Byer , Stasko, John T. , Sunderam, Vaidy

This article motivates and describes the dual timestamping methodology, a novel monitoring technique whose goal is to provide first class support for the visualization and animation of distributed and parallel applications. Central to this methodology is the use of both a primary and secondary timestamp in trace events. The primary timestamp is a logical timestamp that provides information about the concurrency of events. This information is useful for generating visualizations that depict the events as occurring in parallel. The secondary timestamp provides a normalized, causality preserving, real-time clock for use in performance visualization. The dual timestamping methodology is the basis for PVaniM, a collection of general purpose and application-specific visualizations of PVM applications. The implementation of PVaniM relies solely on macros and postprocessors. Because system modifications were not required, the PVaniM implementation strategies are general and easily adaptable to other distributed computing system domains.

View