Abstract. Tool support is an important factor for efficient development of parallel programs. Due to different goals, target systems, and levels of abstraction, many specialized tools and environments have been developed. A contrary approach in the area of parallel program analysis is offered by DeWiz, which focuses on unified analysis functionality based on the event graph model. The desired analysis tasks are formulated as a set of graph filtering and transformation operations, which are implemented as independent analysis modules. The concrete analysis strategy is composed by placing these modules on arbitrary networked machines, arranging and interconnecting them. The resulting DeWiz system processes the data stream to extract useful information for program analysis. 1

established in Spring 2004. However it is still possible to generate a nonconformant document if the instructions in the class file documentation are not followed! Be sure to refer to the published Graduate School guidelines at

Abstract. In the recent past, multiple techniques and tools have been proposed and contributed to improve the distributed debugging functionalities, in several distinct aspects, such as handling the non-determinism, allowing cyclic interactive debugging of parallel programs, and providing more user-friendly interfaces. However, most of these tools are tied to a specific programming language and provide rigid graphical user interfaces. So they cannot easily adapt to support distinct abstraction levels or user interfaces. They also don’t provide adequate support for cooperation with other tools in a software engineering environment. In this paper we discuss several dimensions which may contribute to develop more flexible distributed debuggers. We describe Fiddle, a distributed debugging tool which aims at overcoming some of the above limitations. 1

Abstract. Recently the need for extreme scale computing solutions presents demands for powerful and easy to use performance visualization tools. This paper presents a review of existing research on performance visualization for large-scale systems. A general approach to performance visualization is introduced in relation to performance analysis, and issues that need to be addressed throughout the performance visualization process are summarized. Then visualization techniques from 21 performance visualization systems are reviewed and discussed, with the hope of shedding light on the design of visualization tools for ultra-large systems.

Large scale distributed systems are composed of many thousands of computing units. Today’s examples of such systems are grid, volunteer and cloud computing platforms. Generally, their analyses are done through monitoring tools that gather resource information like processor or network utilization, providing high-level statistics and basic resource usage traces. Such approaches are recognized as rather scalable but are unfortunately often insufficient to detect or fully understand unexpected behavior. In this paper, we investigate the use of more detailed tracing techniques –commonly used in parallel computing – in distributed systems. Finely analyzing the behavior of such systems comprising thousands of resources over several months may seem infeasible. Yet, we show that the resulting trace can be analyzed using tools that enable to easily zoom in and out on selected area of space and time. We use the BOINC volunteer computing system as a basis of this study. Since detailed activity traces of the BOINC clients are not available yet, we rely instead on traces obtained through a BOINC simulator developed with the SimGrid toolkit and which uses as input real availability trace files from the Seti@Home BOINC project. We show that the analysis of such detailed resource utilization traces provides several non-trivial insights about the whole system and enables the discovery of unexpected behavior.