Observability company groundcover Inc. today added Autonomous Infrastructure to its platform in beta to rightsize the processor and memory requests automatically on customers’ Kubernetes workloads based on how each workload behaves in production.

The technology comes from Kubernetes resource optimization startup Wand Cloud Ltd., which groundcover acquired in September.

The release is aimed at the processor and memory requests Kubernetes has engineers set for each workload up front, usually before it has handled any real traffic, and groundcover said most teams seldom return to those numbers. When one is set too low, the first sign is often a service throwing out-of-memory errors. Padding that goes unused ends up on the monthly cloud bill.

Because the groundcover sensor has been collecting production history all along, a customer on a supported version can open Autonomous Infrastructure for the first time without installing anything and find the data the feature needs already in place. Requested processor and memory for each workload appear next to actual usage. Estimated monthly savings sit beside a count of under-provisioned workloads, and a chart tracks over time how much of the potential saving has been realized.

Dollar figures in the beta are based on a price per processor and memory unit that customers enter themselves. Public cloud pricing is the next step, according to groundcover, and billing data such as the Cost and Usage Report from Amazon Web Services Inc. will come after that.

Most resource optimization tools size each workload in isolation, groundcover argues, so a request that looks right for one service can still starve its neighbors on a crowded node. Stability is the first constraint in groundcover’s own software. Before Autonomous Infrastructure raises a request, it checks how tightly the node is already packed so one workload’s increase does not leave the rest short of resources.

Shahar Azulay, groundcover’s co-founder and chief executive, said customers can check what their clusters stand to save, from data already held in their own cloud accounts, before they let Autonomous Infrastructure “change anything.” The feature stays off until an engineer turns it on and picks the deployments, namespaces or clusters it manages.

During the beta the software touches requests only and leaves limits where they are. Each policy carries minimum and maximum bounds for processor and memory, and an audit log records who assigned each workload to a policy and when. Groundcover suggests starting with lower-risk workloads such as development namespaces and leaving alone something like a production database that needs room to grow.

The beta is available now at no additional cost to every groundcover customer running its sensor on Kubernetes, and the company is also recruiting design partners to work with its product team while it runs. Demonstrations are scheduled for KubeCon + CloudNativeCon North America in Salt Lake City Nov. 9-12.

Groundcover is a venture capital-backed startup that has raised $160 million to date, including a $100 million Series C round led by One Peak in July.

Image: groundcover