Metric analytics and overlays provide percentile calculations and anomaly detection to performance analysis and visualization. You can also see the distribution of metrics and anomalies for a specific time period.
Percentiles are a statistical approach to compare data points, and provide a better way to understand performance trends and whether service-level objectives (SLOs) are being met. The anomaly overlay feature calculates a baseline for a single metric and highlights anomalies on the chart for quick identification.
The metric analytics and overlay features provide the following benefits:
Analytics | Description |
---|---|
Anomaly detection | Calculates the baseline and highlights the data that exceeds three standard deviations. |
Custom percentile aggregation | Calculate percentile aggregations for any metric, with the ability to define custom percentiles up to two decimal places. Tanzu Platform collects data at specific intervals by our various collectors and agents. The percentile calculations are based on the data we store. Pre-defined standard percentile options are available for quick selection: P50, P75, P90, P95, and P99. |
Single and multi-time series data support | Calculate and display percentiles for both single-series and multi-time series data, enabling you to compare percentile distributions across multiple datasets. |
Flexible time frame selection | Select a time frame over which they want to calculate percentiles, giving them flexibility in analyzing trends and performance over different periods. |
Visualization options | Select up to 2 percentiles for visualization on any specific widget, enabling you to easily view data against key performance thresholds. |
You can view analytics on any metric widget available on Tanzu Platform for applications, Cloud Foundry, and runbooks.
For example, to view analytics on a metric widget:
Select up to two analytics options from the list.
The metrics widget updates with the selected analytics overlaid on the time series data.