The cloudcompute computational engine is a framework that manages remote execution of containers (referred to as plugins). Cloudcompute relies on regularized structures that declare of inputs and outputs for a container and how inputs and outputs for a series of plugin relate. A cloudcompute simulation consists of a list of linked plugins represented by a directed acyclic graph (DAG) which defines the computational sequence of the plugins for a given event. Cloudcompute also allows for multiple executions of that DAG by allowing the specification of many events to be computed. Cloudcompute allows for events to be run in parallel as well as nodes within the DAG (as far as the DAG will allow) to be computed in parallel.
A plugin is a central idea to cloudcompute. A plugin simply allows for an externally generated software package to be integrated into cloudcompute. The cloudcompute plugins take an input Payload or set of environment variables that define the inputs necessary for execution. A plugin typically is some process model which cloudcompute runs once all inputs have been generated by nodes above the plugin in the DAG. A plugin could be as simple as the generation of a set of random numbers for other plugins to consume, or as complex as a dynamic time-step coupled physics based watershed hydraulics model.
The software development kit (SDK) provides the essential data structures and a handful of utility services to provide the necessary consistency needed for cloudcompute.
GO : https://github.com/USACE/cc-go-sdk
Jave : https://github.com/USACE/cc-java-sdk