Kubernetes has become the backbone of modern infrastructure, powering everything from small clusters at startups to the largest enterprise environments in the world. However, with this scale comes new challenges: rising costs, increasingly complex multi-cloud and hybrid environments, and the constant pressure to use resources more efficiently.
IBM Kubecost has emerged as the standard for monitoring and optimizing Kubernetes spending, providing teams the visibility and insights they need to run efficiently. Over the years, we’ve helped engineering platforms and FinOps teams understand where their money is going, uncover savings opportunities, and take action with confidence.
With IBM Kubecost 3.0, we’re taking a major step forward. This release delivers foundational improvements in performance, scalability, optimization, and automation, while also laying the groundwork for closer alignment between IBM Kubecost and IBM Cloudability. Together, these changes strengthen the foundation for how teams will manage Kubernetes costs in the years ahead.
Platform and engineering teams today face a growing set of challenges:
Kubecost 3.0 is designed to meet these challenges head-on. A new ClickHouse database empowers teams to shift from reactive cost management to proactive optimization by delivering dramatically faster queries, more precise right-sizing insights, and dependable automation that scales effectively across hybrid and multi-cloud setups.
Kubecost 3.0 introduces a re-architected foundation designed for the scale of modern Kubernetes environments. The most impactful change is the migration to a ClickHouse database, replacing DuckDB from version 2.8. With ClickHouse, queries are dramatically faster and more reliable at scale. In internal benchmarks, both allocation and cloud cost API queries saw substantial speed improvements over DuckDB.
The release also debuts a unified agent that combines Kubecost and Cloudability, eliminating the dependency on Prometheus. By removing Prometheus, we’ve substantially reduced the memory footprint, while still supporting open standards through OpenCost. In practice, this means teams can deploy Kubecost more easily with less resource overhead.
Finally, we’ve introduced a simplified topology where each pod runs a single container. This design makes it easier to scale components independently, improves system resiliency, and simplifies onboarding and troubleshooting. The result is a cleaner, more maintainable architecture that reduces time-to-resolution when issues arise.
Read the full architecture migration guide
Kubecost 3.0 introduces major upgrades to Kubernetes optimization. From container requests to node groups, this release makes right-sizing more actionable, easier to automate, and better suited to complex, dynamic environments.
Automated Container Request Sizing introduces a powerful new UI to manage requests across all clusters, enabling teams to start saving resources with just a few clicks. The experience is multi-cluster aware and supports custom profiles, with the ability to adjust CPU and RAM independently. Teams can suspend actions when needed, audit completed resize runs and compare recommendations against realized savings. This capability is powered by v2 licensing, which enables secure messaging between clusters. It’s even available on the free tier (limited to 250 cores) and optimized for EKS primary clusters.
Learn more about Automated Container Request Sizing
Container Request Sizing Insights have also been expanded. New usage visualizations make recommendations easier to interpret, and engineers can now archive them for historical reference. Reports can be saved and exported more flexibly, with enhanced CSV outputs that include labels and a new PDF export option. These improvements give teams more flexibility in acting on Kubecost recommendations and sharing results internally.
Learn more about Container Request Sizing Insights
One level up, Node Group Sizing (formerly Cluster Right-Sizing) has been completely revamped in 3.0. Recommendations are GPU-aware and presented with redesigned usage and capacity visualizations that make it easier to interpret what’s driving inefficiency.
Savings and cost calculations are also more realistic, thanks to reconciled billing data that accounts for customer-specific discounts like Enterprise Discount Programs (EDPs) and Reserved Instances (RIs). This ensures recommendations reflect not only technical efficiency but also a more accurate representation of the estimated savings available.
Kubecost 3.0 also introduces quantile-based controls, which allow teams to fine-tune recommendations for their own environments. This makes the system more configurable and better suited for complex or highly dynamic workloads, where a one-size-fits-all recommendation isn’t enough. Node group sizing also collapses the Underutilized Nodes card from Savings Insights, consolidating efforts in this area. Finally, the redesigned node group sizing experience returns recommendations faster and with higher fidelity.
Learn more about Node Group Sizing
Kubecost is most valuable when insights are easy to find and act on. In 3.0, we’ve focused on making everyday workflows smoother, reducing noise, and surfacing the information teams need without extra steps.
Advanced Filters in 3.0 allow teams to build complex “AND/OR” logic directly in the UI. This is especially valuable in large, dynamic environments, where teams need to combine multiple dimensions such as namespace, label, and service. Instead of stitching queries together outside the Kubecost UI, users can now create and run these complex filters natively, getting precise answers in just a few clicks.
Learn more about Advanced Filters
Alerts in 3.0 have been restructured to streamline where and how alerting features are managed. Budget alerts now live on the Budget page as configurable actions, and recurring alerts have been replaced with scheduled reports that support auto-send functionality. These changes move core alerting workflows into more purpose-built areas of the product.
What remains on the Alerts page is a new health alert tied directly to the diagnostics page. Its purpose is simple: to notify you whenever Kubecost doesn’t have all the data it needs, so you can trust the data you do see. This provides an immediate signal when something’s missing or broken, along with the context needed to resolve it. A fully redesigned alerts experience is planned for Kubecost 3.1, building on this foundation.
Finally, the Collections landing page has been redesigned to better guide new users. It now prompts teams to start with categories and makes it easier to share Collections, helping organizations collaborate more effectively around Kubecost insights.
Kubecost 3.0 also introduces an update to our free tier that reflects the broader scope of the product. Kubecost is no longer just about Kubernetes. We’ve added more cost monitoring and management capabilities for all your Cloud Service costs as well as OpenCost Plugins. With these new capabilities, teams can manage non-Kubernetes costs with features like collections, budgets, forecasting, and anomaly detection. This makes Kubecost valuable even beyond Kubernetes and helps organizations build a more complete view of their cloud efficiency.
To reflect this expanded functionality, 3.0 introduces a new free tier gate based on total spend. The free tier still covers environments with over $1M in annualized cloud costs, giving most teams plenty of flexibility to adopt Kubecost without hitting usage limits. (The gate is measured as $100,000 USD over the last 30 days.)
Kubecost 3.0 also introduces Telemetry, a new way for us to understand how Kubecost is being used to help improve the product faster. Staying opted-in is one of the best ways users can contribute—telemetry directly shapes the roadmap and helps us identify where to make the next big improvements.
Kubecost 3.0 is more than a feature release—it lays the foundation for where we’re headed. Improvements in performance, scalability, optimization, and automation give teams the tools they need today, while preparing for what’s next.
This release also sets the stage for tighter alignment between Kubecost and Cloudability. Together, the two will provide a more complete view of infrastructure spend by bringing Kubernetes cost visibility into the broader world of cloud cost management. For teams operating in hybrid or multi-cloud environments, this convergence enables side-by-side management of container and non-Kubernetes cloud costs with the same level of trust and precision.
Looking ahead, Kubecost will continue to build on this foundation with deeper automation, stronger integrations, and a fully reimagined alerts experience in 3.1 and beyond.
Kubecost 3.0 is available today. Here are a few ways to get started:
We’re excited to see how you use 3.0 to get more from your Kubernetes and cloud environments. Stay opted in to telemetry to help shape the roadmap and keep an eye out for more innovation and integrations in the releases ahead.
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。
