Operator lifecycle manager operator memory is spiking up as we increase the number of namespaces at scale - 750 MB to 13 GB at 500 node scale and 11.6 GB at 250 nodes: https://bugzilla.redhat.com/show_bug.cgi?id=1924973. Discussing with the OLM team ( dsover-1  and bluddy ), they suggested that one of the improvements can be to improve the CSVs creation/handling as namespaces grow. This card is meant to discuss/track all the enhancements needed to improve OLM resource usage on a large and dense cluster.