It would be useful for the collector to be able to filter the collected data, particularly when using backends such as Prometheus that struggle with high cardinality data sets.
An example would be to limit the metrics from dataset 0 (the System dataset) so that we only collect for a limited number of names.
A similar but different enhancement would be to add the ability to e.g. add threshold-based filtering where we would avoid recording the "noise" (again, useful if there's one heavy user and we could avoid recording all of the others thereby avoiding the cardinality explosion if the total user count is high).
It would be useful for the collector to be able to filter the collected data, particularly when using backends such as Prometheus that struggle with high cardinality data sets.
An example would be to limit the metrics from dataset 0 (the System dataset) so that we only collect for a limited number of names.
A similar but different enhancement would be to add the ability to e.g. add threshold-based filtering where we would avoid recording the "noise" (again, useful if there's one heavy user and we could avoid recording all of the others thereby avoiding the cardinality explosion if the total user count is high).