next up previous index
Next: 7.2 Splitting and Data Up: 7.1 Simulation Data Previous: 7.1.4 Meta Data

7.1.5 Persistence and Robustness

While in principle all data generated by simulation tools is completely derivable from a simulator-specific flow description, computation costs (in terms of time and efforts) demand that all results be kept accessible from the moment they are produced on until they are explicitly removed. This holds for mere sequential operation as well as for parallel computation of split-lot experiments. More precisely, it is not acceptable to lose precious data due to malfunctions of a tool, the framework, the operating system, the network, or some other component of the computing environment. In VISTA/SFC, great care has been exerted to ensure a continual update of information stored on the file system to reflect the current state of all tasks in progress.



Christoph Pichler
Thu Mar 13 14:30:47 MET 1997