Hosts reporting high write latency on multiple luns

A FlexVol or FlexGroup is experiencing increased / higher than expected write latency.
Volumes / Luns are spread across both the nodes.
Latency is reported as Data and Disk latency from qos statistics or AIQ
systat-x output shows evidence of CP type CP_Ty B (Back-to-Back) CPs
systat -m output shows CPU more than 90% around the time of the issue:

Note: Columns removed to improve readability

node01::*> node run -node node1 sysstat -m Warning: These diagnostic commands are for use by NetApp personnel only. ANY AVG CPU0 CPU1 CPU2 CPU3 CPU4 CPU5 CPU6 59% 94% 93% 94% 94% 94% 94% 94% 72% 94% 94% 95% 95% 95% 95% 95% 75% 95% 94% 95% 94% 95% 94% 95% 62% 94% 94% 94% 95% 95% 94% 94% 79% 95% 96% 96% 95% 97% 95% 96% 82% 95% 95% 95% 95% 96% 95% 96% 81% 95% 96% 96% 95% 97% 95% 97% 86% 95% 96% 96% 96% 99% 96% 97% 83% 96% 96% 96% 96% 97% 96% 97%