AWS Kubernetes teams struggle to control capacity costs
Teams running AWS Kubernetes clusters through Karpenter want to coordinate Savings Plans, On-Demand and Spot capacity, but instance types, prices and provisioner limits make allocation and spend ceilings difficult to estimate manually. Operators also want different interruption responses for different workloads: early migration can reduce downtime for some apps, while unnecessary node replacement can hurt others. A separate AWS Amplify bill-shock report points to adjacent demand for more predictable cloud costs, though it is not specifically about Karpenter.
For AWS Kubernetes platform and infrastructure teams. Mentioned from Jan 2022 to Nov 2024 on GitHub and Hacker News.
5 different people described this problem in 5 separate discussions.
- Indie fit
- 6.0/10
- Pain
- 7.1/10
- Frequency
- 6.5/10
- Willingness to pay
- 3.8/10
- Momentum
- 5.0/10
- Who pays
- Businesses
- Competition
- High
- Build difficulty
- High
What people said
Quoted word for word. Follow a link to read the whole discussion.
Users have to manually calculate maximum spend by what their provisioner's requirements/limits. And with spot its even less controlled cost
sidewinder12s on GitHub (aws/karpenter-provider-aws)Jan 2023+38 upvotesAsked for a toolHas a workaroundBut then we need to manage multiple deployments, plus with different instance types and prices it's quite a manual process to keep a track of the utilisation of the SP by the "OnDemand" deployment
See what to build and who will buy it
- 2 product ideas with the smallest useful version and pricing
- 4 places to find your first customers
- 5 more quotes from people who have this problem
- Current workarounds, existing solutions and risks