Affinity bottleneck due to a few volumes sharing the same VOL affinity ID

Applies to

• ONTAP 9.3 or later
• WAFL/Volume Affinity Remapping

Issue

• ONTAP isolates volumes to different processing paths called affinities
• The individual volumes only get a subset of the total controller's CPU performance
• High latency can be seen from the graphana tool at customer end for the impacted volume 

• There are a limited number of volume affinites which all volumes share
• A narrow issue exists where two volumes or more may have enough workload on the same affinity to cause resource contention
• This often occurs after an ONTAP reboot, because the Affinity ID is re-randomized during the restart process.