Chiron is a workflow execution engine designed to execute scientific workflows in parallel in High Performance Computing (HPC) environment. A major goal of Chiron is to take a workflow specification and provide for data parallelism automatically with runtime query provenance support. Data is fragmented from a set of parameter sweep combinations or input dataset. Parallel processing is obtained in a MapReduce (Hadoop) style, however, Chiron engine is supported by a workflow algebra, which allows for optimization, dynamic scheduling and runtime workflow steering. Some additional libraries are necessary to execute Chiron such as the JDBC drivers to connect with PostgreSQL database and MPJ libraries.

Chiron Website
Source code is available on SourceForge.


SciCumulus is an encapsulation of Chiron for cloud environments. In other words, it is cloud workflow engine. SciCumulus supports parallel execution of scientific workflows using the Relational Algebra implemented on Chiron. This engine also supports retrospective provenance, generating runtime provenance data (information about the workflow execution). Furthermore, SciCumulus is responsible for automatically adapting the workflow execution according to the current state of the environment. Thus, it is benefited by elasticity and adaptivity.

More information and source code are available on SourceForge.