working-with-intervals

You can filter time-bounded data by duration windows or time ranges using the filter command in OPAL. After computing a duration column, apply a filter like "dur > 5m" to isolate long intervals, enabling you to identify slow CI/CD runs or throughput bottlenecks.

Intervals represent time-bounded processes with both start_time and end_time, whereas events are point-in-time occurrences and resources are static entities. Analyzing interval datasets focuses on measuring durations and distributions, which is essential for evaluating time-bounded activity like batch jobs and distributed traces.

Yes, you can group duration statistics by fields like service, host, or dataset using the statsby command in OPAL. This allows you to aggregate interval data and compare percentile-based statistics, such as p95 duration, across different grouping fields within your time-bounded datasets.

The best way to analyze distributed traces for slow intervals is to treat them as two-timestamp interval datasets and compute duration distributions. By applying percentile-based statistics and duration filters in OPAL, you can accurately surface and identify slow time-bounded activity patterns.

No specific dependencies are required to compute percentile statistics for batch jobs. You can directly analyze interval datasets in OPAL using built-in commands like make_col for durations and statsby with percentile functions to evaluate time-bounded processes without external components.