Section 4 describes the construction of the O&M knowledge graph for an MSA application in theKubernetes environment.3.2.2. Causality Graph ConstructionThe causal analysis method proposed in this paper constructs a causality graph with KPI data.The definition of causality is that given two performance indicators X and Y, if performance indicatorX affects another performance indicator Y, it is denoted as a directed edge X → Y, or we say X causedY; in other words, X is a parent of Y. If we can clarify the causality relation between two performanceindicators X and Y, but it is not clear whether X affects Y or Y affects X, then it could be denotedby X X Y with undirected edges. Note that, in the causality graph, there is no circuit between twoindicators, which means they cause each other. Thus, an ideal causality graph is actually a directedacyclic graph (DAG), as in Definition 1.