Cadence and Time Resolution
Mona now allows higher cadence and more granular time resolutions for insight generation, along with a “priority” channel for separating batch and real-time processes.
In some cases, users might need to track their system’s performance at higher time resolutions in order to catch anomalies more quickly. For this, Mona now allows you to set lower thresholds for both the cadence of the insight generation, and the time_resolution of insights per verse or stanza.
NOTE: these features are currently in beta so please contact us before setting these parameters.
The cadence parameter can be applied to all verse types and it will determine the cadence in which the insight generator will run.
The time_resolution parameter tells your verse on what resolution it should search for anomalies. This currently affects the anomaly detection only in suddenChange verses and in other verses it only affects the visualization of the time_series.
SuddenChange verses: While the “time_series_points” param sets the number of points in the time series to search for suddenChanges in, the “time_resolution” param will determine the resolution of those points. For example, with “time_series_points” : 60 and “time_resolution” : “1h”, Mona will look for suddenChange in the last hour in comparison to the 59 hours prior to it.
Other verses: In other verses this param sets the visualization of the time series but does not affect the anomaly detection.
The expected format for both params is "" where NUMBER is a positive integer, and RESOLUTION options currently include: "h" (hours), "d" (days), or "w" (weeks). e.g, "1d" means 1 day period.
The default value for both cadence and time_resolution is “1d” but can be changed to these following values:
Minutes: 15m, 20m, 30m.
Hours: 1h, 2h, 3h, 4h, 6h, 8h, 12h.
Days: 1d, 2d, 3d, 4d, 5d, 6d.
Weeks: 1w, 2w, 3w, 4w, 5w.
In order to support fast data ingestion for higher cadence insights, we also support an ingestion channel for priority data separated from the regular data ingestion channel (which may be used for batch processes as well), which we can activate upon request.