Zesty

Zesty automates cloud resource optimisation, covering Kubernetes resource optimisation with multi dimensional autoscaling, adaptive pod placement and persistent volume management, aimed at eliminating waste without degrading performance.

Acting automatically rather than recommending is the distinction that matters, and it addresses why most cost tooling underdelivers. Recommendation engines produce long lists of savings that require an engineer to evaluate, schedule and apply, and those lists sit unactioned because the engineer has features to ship and the saving is somebody else's budget. Software that makes the change itself converts a theoretical saving into a real one.

The reason teams over provision is rational rather than careless: under provisioning causes outages and over provisioning only costs money, so every incentive points at buying headroom. Automation that adjusts continuously can hold a tighter margin than a human would dare, because it responds within minutes rather than at the next review.

Kubernetes is the right place to concentrate. Its resource model asks engineers to declare requests and limits per container, values almost nobody can estimate correctly, so the gap between requested and actually used capacity is where most of the waste in a modern estate sits.