Kubernetes resource management is a critical shared responsibility for platform and application teams. Historically, the priority for most teams has been to ensure that Kubernetes clusters and Kubernetes-deployed applications are stable and performant. As FinOps tooling and awareness have improved, though, cloud cost management has become an equally important priority for organizations optimizing their Kubernetes resource use.
The open Kubernetes ecosystem provides many powerful building blocks to help manage resources effectively, including the native Horizontal Pod Autoscaler, the Vertical Pod Autoscaler, Karpenter, and the Cluster Autoscaler. However, effectively implementing these tools in a way that promotes continuous, cost-efficient performance often requires a significant amount of effort from platform and application teams.
To help solve these challenges, Datadog Kubernetes Autoscaling provides multi-dimensional workload scaling recommendations and automation, enabling teams to deliver cost savings while maintaining performance and stability. Kubernetes Autoscaling also provides cluster scaling observability, combining the context of critical Kubernetes scaling events with node scaling and cluster efficiency metrics.
In this post, we’ll walk through how Datadog Kubernetes Autoscaling can accelerate Kubernetes cost savings by:
With over 65% of Datadog-monitored containers using less than half of their requested CPU and memory, workload rightsizing has the potential to deliver significant cloud cost savings. In order to help you prioritize your cost-optimization efforts across clusters, Kubernetes Autoscaling provides a cluster list that highlights key information about idle resources and their costs. Additionally, the list includes time-series graphs of recent cost trends for each cluster, helping you contextualize these metrics.
Once you have identified a cluster to target for optimization, Kubernetes Autoscaling provides a workload list sorted on idle consumption to help you begin rightsizing.
As soon as you’ve identified a workload to begin optimizing, you can quickly take action, either directly within the Datadog platform or by using GitOps. The Workload panel provides a complete recommendation for the workload based on your Datadog metrics. For deeper insight into your metrics, you can easily drill down into and inspect them at the individual container level.
Once you have tuned your configuration to meet your workload requirements, you can autoscale your clusters directly from the Workload panel, applying the recommendation as a one-time optimization or enabling automation to ensure that the workload stays tuned on an ongoing basis.
Alternatively, you can export the CustomResourceDefinition (CRD) configuration and apply it within your existing GitOps workflows and review processes. These custom resources can then be accessed and monitored directly within the Kubernetes Explorer in Datadog.
After you enable Kubernetes Autoscaling, you can observe the Kubernetes events—overlaid with your critical workload metrics—as the change propagates to ensure your cluster’s performance and health.
As you rightsize the individual workloads in your Kubernetes clusters, the clusters as a whole can become more efficient. To track this progress over time, the Cluster Scaling overview provides both resource efficiency and node scaling metrics for the entire cluster. As with the workload metrics, you can easily overlay your cluster metrics with your critical autoscaling Kubernetes events, helping you assess the impact of your rightsizing efforts. This enables you to track any efficiency gains and idle cost savings resulting from your autoscaling program for each cluster.
Datadog Kubernetes Autoscaling makes it easy to optimize your Kubernetes clusters based on resource usage, helping you save costs without sacrificing performance. Detailed workload metrics give you granular visibility into cluster activity, while long-term cost data helps you track the impact of rightsizing these workloads.
Kubernetes Autoscaling is generally available—you can get started using our documentation. Or, if you’re not yet a Datadog user, you can get started with a 14-day free trial.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。