Kubecost Launches Open Source Project to Rein in K8s Costs
A tool for monitoring and optimizing spending on Kubernetes clusters developed by Kubecost has become an open source project.
Alex Thilen, head of business development at Kubecost, says OpenCost is now available as open source software that the company has already submitted to the Cloud Native Computing Foundation (CNCF) for approval as a sandbox-level project. In addition to Kubecost, founding members of the project include Adobe, Armory, Amazon Web Services (AWS), D2iQ, Google, Mincurv, New Relic and SUSE.
The OpenCost community will develop industry standards and vendor-neutral documentation to promote a set of methodologies and best practices for tracking Kubernetes costs regardless of which distribution is employed, says Thilen.
Thilen says it’s the right time to launch OpenCost because more Kubernetes management platforms are providing cost controls. It doesn’t make sense for each platform to reinvent the same basic capabilities that Kubecost already provides, he notes.
Kubecost was originally developed by Stackwatch. Kubecost subsequently raised $25 million to build tools and applications on top of what has now become OpenCost.
OpenCost is designed to run inside a Kubernetes cluster, but no data is sent out of a cluster without user permission. It can collect data in real-time after being installed, which only takes a few minutes.
The primary issue that OpenCost addresses is overprovisioning of Kubernetes infrastructure. Out of an abundance of caution, many developers will overprovision infrastructure to ensure maximum application performance, The issue that creates is that much of that infrastructure goes unused; costs rise steadily as each new Kubernetes cluster is provisioned. Kubecost estimates organizations can reduce Kubernetes-related cloud spending by 60–80% without impacting application performance.
Of course, many enterprise IT organizations will have signed contracts with cloud service providers that guarantee discounted pricing if they run a certain number of workloads per month. However, many IT organizations still prefer to continuously monitor pricing offered by multiple cloud service providers to reduce their overall costs. Regardless of approach, interest in reining in costs is on the rise as the percentage of workloads running on cloud platforms continues to steadily increase.
As the percentage of workloads running on Kubernetes clusters steadily increases, the more likely it becomes those platforms will be centrally managed by an IT operations team. Those teams are evaluated on how well they optimize usage of cloud infrastructure. Those teams also need to show development teams how much Kubernetes infrastructure is actually being consumed by individual applications.
Changing economic conditions also mean there is a lot more focus on cost control today than there was just a few months ago, notes Thilen. Finance teams, especially, are asking harder questions about IT spending than they tended to do at the start of the COVID-19 pandemic when the primary focus was shifting as many workloads to the cloud as possible.
The primary issue organizations encounter in reining in cloud costs is simply the lack of visibility that OpenCost can provide, adds Thilen. It’s not clear when Kubernetes cost controls might be pervasively included in every management tool, but the existence of OpenCost suggests it’s now a matter of when that will occur rather than if those tools will become much more accessible.
