Cesare Pautasso, Win Bausch, Gustavo Alonso
Jürg Kohlas, Bertrand Meyer, André Schiper (eds.)
Dependable Systems: Software, Computing, Networks, Springer, pp. 211-230
Virtual laboratories can be characterized by their long-lasting, large-scale computations, where a collection of heterogeneous tools is integrated into data processing pipelines. Such virtual experiments are typically modeled as scientific workflows in order to guarantee their reproduceability. In this chapter we present JOpera, one of the first autonomic infrastructures for managing virtual laboratories. JOpera provides a sophisticated Eclipse-based graphical environment to design, monitor and debug distributed computations at a high level of abstraction. The chapter describes the architecture of the workflow execution environment, emphasizing its support for the integration of heterogeneous tools and evaluating its autonomic capabilities, both in terms of reliable execution (self-healing) and automatic performance optimization (self-tuning).
PDF: ▼jopera-hasler2006.pdf (2MB)