Pattern similarity detection for cloud resource anomaly analysis

A system for analyzing resource utilization in at least one computing cloud, comprising:

• at least one processor executing a code for:
• receiving a target resource utilization selected from a plurality of resource utilizations of at least one computing cloud, by at least one of via a graphical user interface (GUI) and automatic selection,
• wherein the target resource utilization is automatically selected in response to automatically detecting an anomaly of resource utilization that exceeds a predicted value or exceeds a predefined value;
• generating a target delta time-sequence by computing a plurality of deltas between a plurality of time-spaced values of the target resource utilization over a time interval;
• computing a target vector representation of the target delta time-sequence;
• generating a plurality of candidate delta time-sequence for each of a plurality of candidate resource utilizations of the at least one computing cloud, each candidate delta time-sequence computed as a plurality of deltas between a plurality of time-spaced values of the candidate resource utilization over the time interval;
• computing a plurality of candidate vector representations of the plurality of candidate delta time-sequences,
• wherein each of the target vector representation and the plurality of candidate vector representations denote a rate of change and/or a direction of change of the corresponding time-space values of the corresponding resource utilization;
• computing a plurality of similarity measures, each similarity measure between the target vector representation and a respective candidate vector representation;
• selecting at least one of the candidate vector representations associated with the similarity measure meeting a requirement;
• and
• presenting on a display and within the GUI, at least one candidate resource utilization of the plurality of candidate resource utilizations corresponding to the selected at least one of the candidate vector representations.

A non-transitory medium storing program instructions for analyzing resource utilization in at least one computing cloud, comprising program instructions which when executed by at least one processor, cause the at least one processor to:

• receive a target resource utilization selected from a plurality of resource utilizations of at least one computing cloud, by at least one of via a graphical user interface (GUI) and automatic selection,
• wherein the target resource utilization is automatically selected in response to automatically detecting an anomaly of resource utilization that exceeds a predicted value or exceeds a predefined value;
• generate a target delta time-sequence by computing a plurality of deltas between a plurality of time-spaced values of the target resource utilization over a time interval;
• compute a target vector representation of the target delta time-sequence;
• generate a plurality of candidate delta time-sequence for each of a plurality of candidate resource utilizations of the at least one computing cloud, each candidate delta time-sequence computed as a plurality of deltas between a plurality of time-spaced values of the candidate resource utilization over the time interval;
• compute a plurality of candidate vector representations of the plurality of candidate delta time-sequences,
• wherein each of the target vector representation and the plurality of candidate vector representations denote a rate of change and/or a direction of change of the corresponding time-space values of the corresponding resource utilization;
• compute a plurality of similarity measures, each similarity measure between the target vector representation and a respective candidate vector representation;
• select at least one of the candidate vector representations associated with the similarity measure meeting a requirement;
• and
• present on a display and within the GUI, at least one candidate resource utilization of the plurality of candidate resource utilizations corresponding to the selected at least one of the candidate vector representations.