Currently the maximum number of Node Pools on the AKS Cluster is 100, however we would like to be able to have more then 100 Node Pools.
On our environment we use node pools to isolate different types of workloads onto their own nodes to limit the noisy neighbour effect. When using persistent volumes, it is recommended to have separate node pools per availability zone. Furthermore, during upgrades we prefer to create replacement node pools instead of upgrading the existing ones (blue green deployment). The reason we chose b/g upgrades for certain nodepools is that some of our applications are not fully cloud-native and don't play well with draining of nodes. These applications require special care and manual intervention when rescheduling pods to different nodes.
For these cases we prefer b/g upgrades over in-place upgrades.
This leads to the following calculation:
Number of workload types Number of availability zones 2 (for upgrades) = Number of needed nodepools
Currently the maximum number of Node Pools on the AKS Cluster is 100, however we would like to be able to have more then 100 Node Pools.
On our environment we use node pools to isolate different types of workloads onto their own nodes to limit the noisy neighbour effect. When using persistent volumes, it is recommended to have separate node pools per availability zone. Furthermore, during upgrades we prefer to create replacement node pools instead of upgrading the existing ones (blue green deployment). The reason we chose b/g upgrades for certain nodepools is that some of our applications are not fully cloud-native and don't play well with draining of nodes. These applications require special care and manual intervention when rescheduling pods to different nodes. For these cases we prefer b/g upgrades over in-place upgrades.
This leads to the following calculation:
Number of workload types Number of availability zones 2 (for upgrades) = Number of needed nodepools
Thanks