Product2 min read
ByEmmanuel Bakare
Which node autoscaler should you run?
Compare autoscaler costs and see what you could save by tuning your setup or switching.
Your node autoscaler affects how much compute your cluster runs and which capacity it buys. As workloads change, it's worth checking whether the same setup still fits.
We've added an autoscaler comparison to CostGraph so you can review the estimated cost of supported options. It separates savings you can get by tuning your current setup from the additional savings of switching.
Each option shows its projected monthly cost and saving against today. The breakdown separates packing, capacity, and fee savings from what tuning alone would recover, and the checks list what to review before you move.
Compare projected costs
CostGraph estimates the nodes needed for the same workloads at their recommended requests, using your cluster's current cost as the baseline. The comparison shows how tighter packing and a different mix of instance types or spot capacity affect projected monthly cost.
Savings from tuning your current autoscaler appear separately. This helps you decide whether switching adds enough value to justify the migration.
Supported comparisons
Options include Karpenter, managed node groups with Cluster Autoscaler, and cloud-managed provisioning on supported AWS, Google Cloud, and Azure clusters. Available comparisons depend on your provider and pricing coverage. Options with unpriced management fees are listed without a cost comparison.
Review migration requirements
The recommendation includes checks for disruption budgets, spot interruption handling, and platform constraints. Requirements that cannot be checked are flagged for review.
If you decide to switch, plan a staged migration. Run the new autoscaler alongside the current one, move workloads gradually, and remove the old setup after verifying the move. Review any proposed infrastructure changes before applying them.
Book a demo and we'll compare the options for your clusters with you.