Achieving a rolled-up view of SKA TM health status and state: definition and analysis of aggregation methods

摘要

One basic system requirement for the SKA Project is aimed to providing both a rolled-up and a drilled-down view of the Operational (Health) Status, as well as the State, for each Element. The first one is a performance indicator that must be properly structured in order an operator can understand it. The second one is a logic enumerative that has to contain at least the following values: start-up, shutdown, standby and operational. In this paper an aggregation algorithm, defined for the Health Status of the SKA Telescope Manager (TM) but abstract enough in order to be extended to a generic complex system, is analyzed in terms of the three main indications it has to provide: fully operating, degraded performance, faulty. The analysis shows that the distinction among critical and non-critical components, along with their state and weight on the overall system, is enough to completely achieve a rolled-up view of the system. System architecture only affects the weights, which are the result of a dependability analysis initially performed on the system itself. The aggregation can be applied to selected subsets of components, so providing hints for a drilling-down of the monitoring data, when this is necessary. Finally, using the algorithm to aggregate the states essentially fails: a different approach based on logic relationships will have to be adopted in order to manage states, but to this purpose an important role can be played by the aggregated health status.