Metrics and Monitoring

Metrics and monitoring are the two analysis layers of the platform. They work together: metrics transform events into time series, and monitors observe those series to detect behaviors outside the expected range.

How they relate

Everything begins with the events your operation sends. From those, you configure a metric to aggregate that data over time — counting occurrences, summing values, calculating rates. The metric is the lens through which you see what is happening.

With the metric in place, you configure a monitor to watch it automatically. The monitor runs at regular intervals and opens an alert when it detects something outside the normal pattern — either by a manual rule you defined, or by an anomaly identified through machine learning.

Event Metric Monitor Alert
Metric
type:   "transaction"
agg:    count
dim:    "status"
filter: "BRL"
time series
observes
Monitor
metric:    "transaction"
interval: 5min
rule:     anomaly
alert:    ↑ spike
continuous surveillance

When to use each

Use metrics to answer questions about the past and present: how many transactions per hour? What is the average approved value? The metric is a tool for observation and analysis.

Use monitors when you need to be automatically notified about relevant changes: a sharp drop in volume, an abnormal increase in declines, a value that exceeded a critical threshold. Monitoring is the continuous surveillance layer.


Table of contents