LaunchMON: An Infrastructue for Large Scale Tool Daemon Launching

LaunchMON is a tool infrastructure that allows a HPC tool to deploy tool daemons into the right remote nodes. It makes use of a target resource manager (RM)'s Automatic Process Acquisition Interface (APAI) to identify the remote nodes and processes of a parallel program, and also exploits the same RM's efficient MPI job launching capability on co-locating daemons with the job. To support a wide range of HPC tools that are inherently distributed software, LaunchMON provides the tool with distributed application programming interface sets: the front end (FE) API, the back end (BE) API and the middleware (MW) API. They each support a tool's front end, back end daemons and middleware communication daemons, respectively. Using those API sets, the tool can launch and initialize their remote daemons scalably.