Optimizing EKS Resource Management with Goldilocks

In most EKS clusters, the majority of costs stem from the EC2 instances utilized for running containerized workloads. Achieving optimal sizing of compute resources necessitates a deep understanding of workload requirements. Therefore, it’s crucial to appropriately configure requests and limits and make necessary adjustments to these settings. Furthermore, factors like instance size and storage choices can impact workload performance, potentially resulting in unforeseen cost escalations and reliability issues. 

Achieving effective resource management in Amazon Elastic Kubernetes Service (EKS) requires finding a delicate balance. It’s imperative to allocate sufficient resources (CPU, memory) to ensure smooth functioning of pods, yet excessive provisioning can lead to resource wastage and increased expenses. This is where the concept of Goldilocks comes into play, striving to identify the ideal resource allocation that is neither too much nor too little for your EKS deployments. 

What is Goldilocks? 

Goldilocks is an open-source tool specifically designed to help optimize resource requests and limits in Kubernetes environments like EKS. It works by analyzing historical resource usage data of your pods and recommends optimal resource allocations. This helps you avoid: 

▸ Over-provisioning: Wasting money by allocating more resources than your pods need. 

▸ Under-provisioning: Causing performance issues for your pods due to insufficient resources. 

Problem Statement: 

Managing resources effectively in Kubernetes clusters, especially in managed environments like Amazon Elastic Kubernetes Service (EKS), presents challenges related to resource allocation, optimization, and cost control. Without granular insights into resource utilization and performance, organizations face the following challenges: 

1. Resource Overprovisioning

▸ Without visibility into actual resource usage patterns, teams tend to overprovision resources, leading to wasted capacity and increased operational costs. 

▸ Overprovisioning also reduces cluster efficiency and scalability, as resources are not utilized optimally. 

2. Underutilization and Performance Issues

▸ On the other part, underutilization of resources can result in performance issues, such as slow response times and degraded application performance. 

▸ Without real-time monitoring and analysis, it’s challenging to identify underutilized resources and optimize resource allocation accordingly. 

3. Cost Inefficiency

▸ Inefficient resource allocation leads to unnecessary spending on cloud resources, impacting the organization’s budget and overall cost-effectiveness. 

▸ Without cost optimization measures, organizations struggle to control cloud infrastructure expenses and allocate resources judiciously. 

Solution: Implement goldilocks on EKS 

Goldilocks addresses above challenges by providing intelligent resource allocation recommendations based on real-time usage metrics and workload characteristics. Here’s how Goldilocks solves the above problems: 

1. Optimized Resource Allocation:

▸ Goldilocks analyses historical resource usage data and predicts future resource requirements based on workload patterns and resource utilization trends. 

▸ By providing recommendations for resource requests and limits at the pod level, Goldilocks helps organizations optimize resource allocation and right-size Kubernetes deployments, ensuring efficient resource utilization without overprovisioning. 

2. Improved Performance and Scalability:

▸ Goldilocks identifies underutilized resources and suggests adjustments to resource limits to improve application performance and scalability. 

▸ By dynamically adjusting resource limits based on workload demands, Goldilocks ensures that Kubernetes clusters can scale effectively to handle varying workloads without sacrificing performance or reliability. 

3. Cost Optimization:

▸ Goldilocks helps organizations control cloud infrastructure costs by optimizing resource allocation and minimizing waste. 

▸ By providing recommendations for resource requests and limits that balance performance requirements with cost considerations, Goldilocks enables organizations to achieve cost-efficient Kubernetes deployments without compromising on application performance or reliability. 


Goldilocks empowers organizations to optimize resource allocation, improve performance, and control costs in Kubernetes environments running on EKS. By leveraging intelligent recommendations and real-time insights, Goldilocks enables organizations to achieve optimal resource utilization, scalability, and cost-effectiveness in their cloud-native deployments. 

Let’s compare Goldilocks with other tools in the context of Kubernetes resource optimization and management. 

Automation  Goldilocks  Heapster/Metrics Server  Kubernetes HPA  Prometheus with custom alerting rules 
Resource Optimization  ✔️ Intelligent recommendations  Basic resource monitoring  Basic scaling based on metrics  Complex configuration for optimization 
Granular Insights  ✔️ Provides granular insights  Limited visibility into metrics  Limited insight into resource usage  ✔️ Offers comprehensive monitoring and querying capabilities 
Automation  ✔️ Automates resource optimization  Manual configuration required  Manual scaling decisions  Manual configuration for alerting 
Integration  ✔️ Seamless integration with Kubernetes  Native integration with Kubernetes  Native integration with Kubernetes  ✔️ Can be integrated with Kubernetes 
Cost Efficiency  ✔️ Optimizes resource allocation  No focus on cost optimization  Focuses primarily on scaling  No direct focus on cost optimization 


This post explores how Goldilocks helps streamline resource allocation for Kubernetes applications. Modernization projects often face time constraints when determining resource needs. Traditionally, this involves poring over monitoring tools, a cumbersome process. Goldilocks cuts through the complexity by offering data-driven recommendations. By following these suggestions, companies can expedite application launches and reduce costs on Amazon EKS. 


Written by

Mahavishnu Govindaraj

Mahavishnu Govindaraj

Tech Manager - AWS DevOps and Security Specialist

Umashankar N

Umashankar N

Chief Technology Officer (CTO) and AWS Ambassador

Sharing is caring!

In Blog
Subscribe to our Newsletter1CloudHub