High RPM policy shared memory usage

[JaiminLaljibhaiKakadiya9979]

has anyone come across alert on high rpm policy shared memory usage on leaf with high number of contracts. Although from operations tab we do not see any limit being reached on any of the metric (policy cam at 24% and vlan at 52%), customer does make use of lots of global contracts across vrfs and tenants, would like to know what causes high rpm policy shared memory usage and what would help to bring that down

[Robert Burns]

This fault means there is high memory usage caused by Route Policy Map (RPM) policy.
These are usually caused by shared route-map leaks on global VRFs. The issue can be worsened by the reuse of the same contract between all VRFs.  This will result in a high number of route maps applied on leaf switches, so when a new VRF (other than common) becomes a provider, this will cause route leaking between this new VRF and all the other consumer VRFs.
This may warrant a redesign and I'd suggest a TAC case to further investigate/confirm this.

Robert