With extremely large exposition (example: kube-state-metrics on a large cluster will expose hundreds of MiB of metrics data), sorting of metrics becomes a quite significant computational burden. As metrics sorting is not required for valid expositions, a registry could provide an option to not sort after gathering metrics.
With extremely large exposition (example: kube-state-metrics on a large cluster will expose hundreds of MiB of metrics data), sorting of metrics becomes a quite significant computational burden. As metrics sorting is not required for valid expositions, a registry could provide an option to not sort after gathering metrics.