Abstract: Embodiments relate to finding correlations between periods of performance degradation of an application and failure rates of the application. Telemetry data reports results of operations performed by the application. The telemetry data is parsed to identify periods of performance degradation of the operations. Performance measures of successful operations during an identified period are aggregated, and performance measures of failed operations during the identified period are aggregated. For the identified periods, conclusions about positive and negative correlation between failures of the operation and degraded performance of the operations are derived based on the aggregate measures of success and failure and based on rates of failure of the operation during the periods.

Abstract: Embodiments described herein provide dynamic thresholds for alerting users of anomalous resource usage of computing resources. The dynamic thresholds are based on the historical behavior of compute metrics (or a time series obtained therefor) associated with the computing resources and a detected seasonality in that time series. Based on characteristics of the time series, a model for generating dynamic thresholds is determined. The dynamic thresholds track the detected seasonality of the compute metrics. As utilization of the computing resources continue, the determined thresholds are applied to the compute metrics. If the determined thresholds are exceeded, an alert indicating an anomalous resource usage is provided to a user. The dynamic threshold may be adjusted (e.g., tightened or relaxed) based on a confidence level of the detected seasonality. This advantageously reduces the number of false alerts.

Abstract: A compute cloud hosts a distributed application and is configured to add or remove instances of the application at datacenters at disperse geographic regions. Operations of the application are captured in a telemetry stream. Each record in the telemetry stream indicates a time, client location, and performance measure for a corresponding client request. After pre-processing such as rounding the times, the telemetry records are passed to a frequent itemset mining algorithm that identifies frequent time-location pairs in the telemetry stream. The frequent time-location pairs are consolidated into encompassing frequent region time-range pairs. An aggregate performance measure is computed from the performance measures of the telemetry records that match a frequent region time-range pair. A recommended region and time for adding or removing instances of the application is computed based on the aggregate performance measure and the region time-range pair.

Abstract: A compute cloud hosts a distributed application and is configured to add or remove instances of the application at datacenters at disperse geographic regions. Operations of the application are captured in a telemetry stream. Each record in the telemetry stream indicates a time, client location, and performance measure for a corresponding client request. After pre-processing such as rounding the times, the telemetry records are passed to a frequent itemset mining algorithm that identifies frequent time-location pairs in the telemetry stream. The frequent time-location pairs are consolidated into encompassing frequent region time-range pairs. An aggregate performance measure is computed from the performance measures of the telemetry records that match a frequent region time-range pair. A recommended region and time for adding or removing instances of the application is computed based on the aggregate performance measure and the region time-range pair.

Abstract: Embodiments relate to finding correlations between periods of performance degradation of an application and failure rates of the application. Telemetry data reports results of operations performed by the application. The telemetry data is parsed to identify periods of performance degradation of the operations. Performance measures of successful operations during an identified period are aggregated, and performance measures of failed operations during the identified period are aggregated. For the identified periods, conclusions about positive and negative correlation between failures of the operation and degraded performance of the operations are derived based on the aggregate measures of success and failure and based on rates of failure of the operation during the periods.