Kubernetes resource management and scheduling are no walk in the park. Accurately forecasting your workload’s needs is complex, leading many teams to under- and over-provisioning, while the scheduler’s decisions may incur unnecessary expenses. Cast AI’s features – Workload Rightsizing & Pod Pinner – solve these challenges with automation and innovative algorithms.
Why is workload rightsizing key for cutting cloud costs?
Workload rightsizing refers to setting your workloads to request the right amount of resources to run smoothly.
In K8s, workloads are rightsized using requests and limits for CPU and memory. This step helps you avoid issues like Pod eviction, CPU starvation, or running OOM. But workload rightsizing is also essential for reducing the cost of running K8s applications.
By limiting your resource usage, you can stop overprovisioning and cut unnecessary expenses.
While Cast AI was able to provide exact workload rightsizing recommendations before, the platform still required engineers to set them manually – until now.
Why automate workload rightsizing?
Automating workload rightsizing is a fast track to optimized resource allocation, cost-efficiency, as well as improved performance and scalability.
By reducing manual tasks, you can save time and effort while ensuring precise adjustments to align your setup with your workload’s actual needs. Moreover, automation helps you avoid human error that could compromise your security and compliance.
Cast AI’s new Workload Rightsizing capability automatically scales your workload requests up and down, ensuring optimal performance and saving you money. It also adds extra overhead to remedy instability if the platform detects any Out-of-Memory container status.
As a result, you can enjoy better performance at a much lower cost – and without adding extra tasks for your engineers.
Workload Rightsizing in real life
Bud enhances financial data by identifying merchant, category, location, and transaction frequency, providing actionable insights and comprehensible inputs for LLMs in the financial services sector.
Using Cast automation, Bud saw a dramatic improvement of CPU and Memory utilization of up to 93%. The team uses Workload Autoscaler to rightsize workloads and reduce CPU requests and unlock new cost savings:
How to use Cast AI’s automated Workload Rightsizing
By default, the platform generates configurable recommendations for each of your workloads every 30 minutes.
You can specify additional overhead for CPU and RAM, adjust their percentile values, and set a threshold for automatically applying the recommendations you get. For example, your recommendations can be applied to the workload only after exceeding certain thresholds.
Cast AI’s team plans to expand automated workload rightsizing further. In the future, the platform will introduce seasonality models for resources to better anticipate hourly, daily, weekly and monthly cycles to improve response time and availability.
Stay tuned, as many exciting developments are coming your way!
Pod Pinner for more efficient K8s scheduling
Alongside the Workload Rightsizing functionality, Cast AI has also introduced Pod Pinner.
This new approach to Kubernetes scheduling focuses on eliminating random pod placement decisions. Powered by an advanced bin-packing algorithm, Pod Pinner ensures strategic pod positioning onto designated nodes to maximize resource utilization while boosting efficiency and predictability across clusters.
Without Pod Pinner, the Kubernetes scheduler might place pods on unintended nodes, resulting in suboptimal resource utilization and higher costs. Moreover, the new feature helps to reduce workload movement to improve both uptime and reliability of workloads and optimize costs along the way.
See automated Workload Rightsizing and Pod Pinner in action
If you’re already using Cast, head to your console and see automated Workload Rightsizing and Pod Pinner in action.
If you are new to Cast AI, book a short technical demo with our engineers to discover what these new features can do for you.
This will be the best-spent 30 minutes this week that will provide you with actionable cost optimization ideas for your K8s cluster!
Kubernetes cost optimization
Monitor organization-wide and cluster-level resource spending. Automate resource allocation and scale instantly with zero downtime.
