Fairwinds has launched an open source tool dubbed Goldilocks designed to make it easier to guarantee and limit the memory resources an application can consume on a Kubernetes cluster.
Robert Brennan, director of open source software for Fairwinds, says that as the IT services provider became more involved in deploying applications on Kubernetes, it became apparent that IT teams needed a tool that could limit the amount of memory any one application could consume on a Kubernetes cluster that is being shared by multiple applications.
Via a Request command, the resources allocated to a container are guaranteed, while a Limits command identifies which resources will be restricted. Kubernetes provides the ability to set default namespace settings for requests and limits; however, to ensure clusters remain stable, Fairwinds advises users of Goldilocks to make sure settings match the unique requirements of each application.
Goldilocks is a dashboard that employs data gathered by a Vertical Pod Autoscaler (VPA) capability in Kubernetes that many providers of such distributions already have available. VPA monitors pods usage to make recommendations for setting resource requests and limits. IT teams can access a service in the cluster and the dashboard will provide two types of recommendations depending on the quality of service required. Kubernetes provides three levels: Guaranteed, which means the application will be scheduled on a node where resources will be assured; Burstable, which means the application will be guaranteed a minimum level of resources and given more if and when available; and Best Effort, which means that no requests or limits are set and the application will be allocated resources only when all other requests are met.
Brennan says the Goldilocks dashboard makes it easier to navigate all those settings for each application running on the Kubernetes clusters.
Fairwinds has launched several open source projects to further ClusterOps, a managed service it provides to outsource the management of Kubernetes platforms. At a time when Kubernetes expertise is hard to find, many organizations are finding outsourcing the management of Kubernetes clusters to be the path of least resistance. That approach enables developers of containerized applications to move forward without having to wait for an internal IT operations team to become certified on Kubernetes.
It’s not clear to what degree Kubernetes clusters will be managed by external service providers versus internal IT teams. However, a recent survey of 750 global senior IT decision-makers at enterprises with 500 to 5,000-plus employees in Australia, France, Germany, the UK and the U.S. conducted on behalf of New Relic, a provider of IT monitoring services, finds more than half of respondents admitted they found their new software and infrastructure difficult to manage and monitor for performance issues.
As the pace at which organizations want to deploy containerized applications continues to increase, however, the need for more Kubernetes expertise is likely to force the issue.