Optimizing Large Grafana Queries

jarurlaj · December 3, 2025, 1:20pm

Hi y’all,

I’m building out a Stat panel in Grafana and was wondering if there is a way to optimize large queries. Currently, the panel is querying 100+ binary points to establish an overall status for the location I’m looking at. Is there a better way to handle this amount of data?

clackner-gpa · December 3, 2025, 10:58pm

It depends a little bit on what you are trying to achieve.

If you are using the latest datasource it will downsample the data by default.

If you are only interested in looking at summaries (like Max, Total, etc) you can use the provided GSF fuctions. Those are evaluated in the openHistorian/openPDC, which reduces the ammount of data Grafana has to handle.

Similarly if you mostly care about a summary of the device you can combine data points using a similar approach (SLICE and SET versions of the GSF functions) that may further rescue the number of tags being send back to Grafana.

A summary of the GSF functions is available here:

github.com

GridProtectionAlliance/gsf/blob/master/Source/Documentation/GrafanaFunctions.md

# GSF Grafana Functions

The Grafana interfaces defined in the Grid Solutions Framework allow for aggregation and operational functions on a per-series and per-group basis. The following defines the [available functions](#available-functions) and [group operations](#group-operations) that are available for a data source implementing the GSF Grafana interface, e.g., [openHistorian](https://github.com/GridProtectionAlliance/openHistorian). Note that any time-series style data source that implements the [GrafanaDataSourceBase](https://github.com/GridProtectionAlliance/gsf/blob/master/Source/Libraries/Adapters/GrafanaAdapters/GrafanaDataSourceBase.cs) class will automatically inherit this functionality.

## Series Functions

Various functions are available that can be applied to each series that come from a specified expression, see full list of [available functions](#available-functions) below. Series expressions can be an individual listing of point tag names, Guid-based signal IDs or measurement keys separated by semi-colons - _or_ - a [filter expression](https://github.com/GridProtectionAlliance/gsf/blob/master/Source/Documentation/FilterExpressions.md) that will select several series at once. Filter expressions and individual points, with or without functions, may be selected simultaneously when separated with semi-colons.

* Example: `PPA:15; STAT:20; SetSum(Count(PPA:8; PPA:9; PPA:10)); FILTER ActiveMeasurements WHERE SignalType IN ('IPHA', 'VPHA'); Range(PPA:99; Sum(FILTER ActiveMeasurements WHERE SignalType = 'FREQ'; STAT:12))`

Many series functions have parameters that can be required or optional &ndash; optional values will always define a default state. Parameter values must be a constant value or, where applicable, a named target available from the expression. Named targets are intended to work with group operations, i.e., [Set](#set) or [Slice](#slice), since group operations provide access to multiple series values from within a single series. The actual value used for a named target parameter will be the first encountered value for the target series &ndash; in the case of slice group operations, this will be the first value encountered in each slice. Named target parameters can optionally specify multiple fall-back series and one final default constant value each separated by a semi-colon to use when the named target series is not available, e.g.: `SliceSubtract(1, T1;T2;5.5, T1;T2;T3)`

To better understand named targets, consider the following steps:

> NOTE: This example is for illustrative purposes only, use the [`Reference`](#reference) function to get a difference between two or more angles taking wrapping and unwrapping into consideration.

 1. The following expression produces two unwrapped voltage phase angle series:

    [`UnwrapAngle(DOM_GPLAINS-BUS1:VH; TVA_SHELBY-BUS1:VH)`](#unwrapangle)

This file has been truncated. show original

jarurlaj · December 10, 2025, 2:10pm

Thanks for reply! I am currently using a SetMaximum function to summarize the points. It is good to know that the function is evaluated on the openHistorian side. Are there any other steps that can be taken?