Letting a provider run the control plane.
WHAT THEY PROVIDE
The control plane operated for you Upgrades Integration with the provider's storage, networking and load balancing
WHAT REMAINS YOURS
Workloads Node configuration, in some offerings Cost management Everything about your applications
WHY THAT DISTINCTION MATTERS
Managed does not mean operated.
WHAT IT COSTS
A charge for the control plane, plus nodes Provider-specific behaviour
WHAT TO CHECK
Which versions are supported, and for how long How upgrades are performed Whether nodes are managed or yours What the networking model is
WHY UPGRADES MATTER
Clusters fall out of support quickly, and upgrading is disruptive.
WHAT TO PLAN
An upgrade cadence, deliberately.
WHAT TO ESTABLISH ABOUT COST
Charges for the control plane, nodes, load balancers, storage and transfer.
WHY TRANSFER SPECIFICALLY
It is frequently the surprise.
WHAT TO WATCH
Nodes running at low utilisation.
WHY
It is the commonest waste in clusters.
WHAT TO CONSIDER LOCALLY
Where the cluster runs, and latency to it Cost in foreign currency
WHAT TO ASK BEFORE ADOPTING ANY OF IT
Whether the workload justifies a cluster at all.
