I have two groups that want metrics in Graphana. One group already has all clusters setup with Prometheus and Graphana. The new group wants to access a subset of these clusters in their current Graphana instance. Would there be any performance issues with the second group standing up their own Harvest pollers or is there a better way to accommodate thier ask?
#Multiple Pollers for a cluster
lethal jasper
mild sand
@lethal jasper Running the same cluster through multiple Harvest instances will increase API load on ONTAP, so that should be avoided. Here are two alternatives:
**Option 1: **Dedicated Prometheus with Selective Scraping (Recommended)
Stand up a separate Prometheus instance for Group 2 that scrapes (same common Harvest instance) only the clusters relevant to their team. This gives them a fully isolated Prometheus and Grafana stack with no additional ONTAP API load and no access to clusters outside their scope.
Option 2: Grafana Variable Filtering
Add a cluster template variable to the dashboards in the existing Grafana instance, pre-scoped to Group 2's clusters. This is quick to implement but offers no true data isolation. Only suitable if there are no security or data boundary requirements between the two groups.
lethal jasper
@mild sand are there any metrics around API load that can be monitored ?
hoary raven
Or you could use NAbox in proxy mode and scrap only the metrics you need