Sign in to my dashboard Create an account

Kubernetes resource optimization in NetApp Cloud Insights

kubernetes resource optimization dashboard graph

Share this page

Miles Kneip
Miles Kniep

We regularly hear from our customers and partners that balancing resource usage in the Kubernetes landscape is both a critical task and—unfortunately—a time-consuming and error-prone endeavor. This is a terrible combination. On the one hand, inefficient use of resources—whether storage, compute power, or network bandwidth—can lead to unexpected and unsustainable increases in TCO. These unanticipated surges can eat into budgets that are allocated for other proactive work, causing unplanned delays and disrupting the smooth functioning of IT operations. On the other hand, being too restrictive with your resource management can lead to performance bottlenecks, system downtime, and, in the worst cases, data loss—all of which can have severe implications for business continuity, customer satisfaction, and, ultimately, the bottom line. So, being able to identify and eliminate resource inefficiencies isn’t just beneficial—it’s business-critical.

To that end, we’re excited to announce the launch of the new container rightsizing feature in our observability platform, NetApp® Cloud Insights. This feature helps you manage your containerized workloads by optimizing resource usage and reducing costs while maintaining the performance and availability you must provide for your end users. Cloud Insights enables your organization to scale key applications to the needs of your business, without the risk of eating up the budget along the way.

Intelligent sizing recommendations

The cornerstone of container rightsizing is its ability to provide intelligent sizing recommendations based on historically observed workload demand. Cloud Insights uses machine learning to understand your workload’s patterns (or “seasonality”) and suggests optimal resource allocation based on several sizing strategies. This approach allows you to adjust your container requests and limits based on actual usage patterns rather than relying on guesswork, static thresholds, or automatic sizing recommendations that are unsuitable for more variable workloads.

kubernetes resource optimization Right sizing Recommendation Memory

CPU and memory consumption analysis

To properly rightsize your workloads in Kubernetes, you need to start by asking the following three key questions. But this kind of analysis soon becomes quite complex if you have to do it manually.

  • How much resource is this container using?
    • How much resource does it typically use?
    • How much resource does it use at peak utilization versus typical usage?
  • How much resource does the container request?
    • Is this in line with expected usage?
  • At what level of usage should I start limiting this container?
    • Will adjusting the limits lead to more frequent interruptions in the application?
    • How can I ensure that the container is reasonably limited, but still strike the proper balance of economy to availability for this workload?

Cloud Insights helps you answer these questions with updated rightsizing guidance as workload behavior is observed over time. This guidance enables you to understand the waste trend across various namespaces and highlights savings potential and at-risk workloads.

kubernetes resource optimization Dashboard

Optimization strategies

Container rightsizing offers three different optimization strategies that we categorize into tiers based on your needs and risk tolerance.

  • Test. This tier is designed to minimize waste and, by extension, your workload’s economic impact as much as possible. It’s appropriate if you’re looking to keep costs as low as possible, recognizing that the workload might occasionally hit its configured limits.
  • Critical. This tier is for you if your priorities are uptime and service continuity. It optimizes workload efficiency but reduces the risk of hitting configured limits, helping avoid potential disruptions or outages.
  • Production. This tier balances the Test tier’s economy and the Critical tier’s availability focus. This balance minimizes the risk of workload throttling or disruption but still provides substantial savings and efficiency.

Gallery dashboard for waste monitoring

With the gallery dashboard provided by Cloud Insights, you can now automatically see the overall waste across your Kubernetes landscape. This visual representation gives you an instant snapshot of resource utilization across clusters. By identifying overprovisioned resources by namespace and workload, you can determine the best opportunities for reclaiming resources.

kubernetes resource optimization Right sizing recommendation table


The container rightsizing feature is a game changer for organizations looking to optimize the TCO of their Kubernetes estate. As we’ve often heard, analysis is typically too time consuming and risky to be worthwhile when left to out-of-the-box tooling or manual efforts. Cloud Insights now simplifies analysis with continuous machine learning and selectable optimization strategies—a combination of power and flexibility.

With Cloud Insights, you can maintain efficient resource use, control costs, and keep your applications performing at their best. We can’t wait for you to try it out— access the free trial now. If you’re curious about what else Cloud Insights can do to help you maintain your service-level objectives on Kubernetes, check out Josh Moore’s blog.

Miles Kniep

Miles is a principal technologist on the Cloud Insights team at NetApp. Having worn many hats in his more than a decade with NetApp, he's worked with countless clients, helping to modernize and enhance their IT Observability practice and helping deliver better outcomes to the business in a myriad of heterogeneous environments. His primary focus is on assisting organizations as they transition to a containerized application landscape to maintain their availability targets while ensuring they do so at a reasonable cost.

View all Posts by Miles Kniep

Next Steps

Drift chat loading