Control Your Kubernetes Costs & Optimize Usage with Rightsizing Recommendations

Apptio Cloudability is the first solution to enable you to broadly apply FinOps principles to Kubernetes, with fully integrated cost management for an accurate chargeback and rightsizing recommendations to help drive efficient usage. Rightsizing for containers is now available to all Cloudability customers, helping maximize the business value realized from Kubernetes investments. By following these recommendations, customers can repurpose freed capacity for additional workloads or reduce the underlying cloud footprint to achieve up to 30% in savings.

Before we dive into how the new container rightsizing capability works, it’s worth spending a few minutes understanding how we got to a world where we need to rightsize yet another set of constructs.

Learning from the recent past – a common set of challenges

Cloudability was developed in 2011 to solve the novel FinOps – or Cloud Financial Management – challenges created by the adoption of public cloud. At the heart of this challenge is how public cloud has transformed the nature of procuring infrastructure. The move from on-premises to public cloud entails shifting away from fixed upfront capital expenditure to variable consumption-based operational expenditure. By completely decentralizing infrastructure procurement, we allowed our delivery teams to innovate quickly, but also contribute to cloud bills that can increase inexorably month to month (or even as quickly as day to day!) with minimal visibility of business benefit.

With this in mind, Cloudability has provided enterprises with a range of capabilities to drive financial accountability and maximize the return on every dollar spent. This starts with harnessing the detailed billing exports from each cloud vendor, grouping costs with native constructs such as resource tags, and applying business rules to ensure all costs are accurately assigned as a part of an automated chargeback. Engineers could then leverage our intelligent rightsizing recommendations to reduce operational costs. These recommendations collect utilization data from native and third-party monitoring tools and typically provide a shortlist of instance types based on official SKUs, aiming to locate an optimal combination of cost and performance.

As it turns out, containerization brings an almost identical set of financial management challenges but requires new methods to source information and models for recommending optimizations. Instead of orienting around regular public cloud interfaces, we need specialist capabilities that read and understand the unique container environment within.

Containerization – A cloud inside a cloud

Containerization is a virtualization technology with several significant benefits over the virtualization that backs public cloud. As a result of virtualizing the operating system layer, (instead of the hardware layer) application developers get benefits both in terms of how software is deployed (lightweight deployment artifacts that can be deployed quickly in different locations) and reduced compute footprint (better bin packing options and sharing an underlying OS). In fact, due to these advantages, according to Gartner, “by 2025, more than 85% of global organizations will be running containerized applications in production, which is a significant increase from fewer than 35% in 2019.”

From a conceptual perspective, we can think of containerized infrastructure as a cloud within a cloud (at least when deployed on a public cloud such as AWS) – the container environment is represented by clusters that are comprised of shared resources (e.g., numerous EC2 instances) and has its own internal resource system and scaling mechanisms. The nature of how these clusters are used means that multiple teams or business units will be responsible for the costs generated across the shared resources.

So how can Cloudability help with the familiar problems of accountability and optimization? Well, in August last year, we launched integrated cost management for Kubernetes, employing sophisticated algorithms to assign costs to Kubernetes constructs – such as Namespaces and Labels – and integrating this detailed allocation information across our analytics platform. With that update, customers can easily include container costs within chargeback processes, manage these costs with official budgets & forecasts, and generally involve them in all aspects of cloud cost management. After making teams accountable, the next logical step is giving them visibility into efficiency and clear options for reducing spend – in this case, recommendations focused on container settings.

Make intelligent trade-offs between cost and performance

We are excited to announce the availability of container rightsizing recommendations. This new capability piggybacks off our previous container releases by leveraging the derived cost contribution for each container and making comparisons between resource settings (Request and Limit) and sub-hourly level K8s utilization data aggregated by Cloudability. As a result, up to three cost-saving options are provided per container for users to evaluate and take action against. The simplest of these is where we detect a completely idle container that you may choose to terminate. Typically, two rightsize options are also provided, which set the Request and Limit for both CPU and memory in different ways. One option sets Request and Limit at an identical level above the previous 10 or 30 days of utilization, surfacing a near-zero risk recommendation. The second option sets Limit above a lower Request value to allow for burst, surfacing a low-risk recommendation. Taking these actions will free up space on existing clusters, allowing operations to repurpose the extra capacity for additional workloads or reduce the underlying cloud footprint to realize up to 30% in savings.

In summary, these rightsizing recommendations enable users to quickly identify cost optimization opportunities within their K8s container environments and maximize the business value driven by K8s, all within the familiar Cloudability rightsizing UI. The engineering team at Apptio recognizes the growing importance of containerization and is committed to continuing our innovations in this area. Stay tuned for more exciting announcements!   

RIghtsizing screen shot - Control Your Kubernetes Costs & Optimize Usage with Rightsizing Recommendations - Apptio
Figure 1: Summary and List of Container Rightsizing Recommendations

To learn more, sign up for a Cloudability free trial today!