| 摘要 |
A method and a network node (110) for localizing a fault causing performance degradation of a service are disclosed. The network node (110) calculates (A020, A030, A040) a first, a second and a third set of correlation values for a first time interval. Similarly, the network node (110) calculates (A070, A080, A090) a fourth, fifth and sixth correlation value for a second time interval in which the fault has been detected. The sets of correlation values are taken between resource metrics, virtual execution environments, types of said each resource metric and/or servers. When said at least one of the absolute values is above the threshold value, the network node (110) adds (A130) said each resource metric to a list of resource metrics, wherein the list indicates resource metric from which the fault potentially originates. Next, the network node (110) sends (A140) the list to an entity (115) configured to manage resources available to the set of servers, thereby enabling the entity (115) to mitigate Service Level Agreement violation. |
