In the Harvest conf file you can define pollers that are of host_type FILER or OCUM. For OCUM it connects to OCUM and collects capacity metrics for all clusters known to OCUM. But, to submit these metrics to Graphite it must know the site for each cluster so it looks for a matching poller name for the cluster of host_type FILER, finds the site, and submits the metrics. If there is no poller entry for the cluster then those metrics are not sent to Graphite.
So if you have an OCUM with 10 clusters and want only perf and capacity info for one cluster you could achieve by putting two poller entries in your Harvest conf file; one for OCUM and one for the cluster you want to monitor. If you have more pairs like this you could simply use multiple Harvest conf files. netapp-manager and netapp-worker have a CLI arg to specify the conf file name (default: netapp-harvest.conf).
One thing to be aware of though is that Graphite, the time-series db, does not support any granular role based access control. So if you want to isolate customer environments you would also need independent Graphite instances. And same for Grafana, it can't restrict data sources, as far as I know. If you run these in containers maybe the manageability is ok...
Hope this helps!
Cheers, Chris Madden
Storage Architect, NetApp EMEA (and author of Harvest)