DSM Ties a Node into a Distributed INtime System

Every node runs DSM, the Distributed System Manager. DSM is a cooperative, distributed process that monitors the system at different granularity: node level and process dependency level.

Node level:

DSM first publishes its node to the other nodes in the system and then actively keeps track of the “health” of all the other nodes. This functionality always is provided, independent of any application running.

Process dependency level:

Processes that need each other to provide full functionality can publish logical dependencies with respective roles, called sponsor and dependent. DSM will monitor these processes and inform the sponsor or dependent in case of normal or abnormal termination.