Kubernetes optimization is easier said than done. The untold reality that we consistently find in the wild is that Kubernetes clusters are massively over-provisioned – average CPU usage at 5% or less – and at the same time, applications are at risk of reliability issues, if not facing actual incidents.

We have built Akamas Insights to help developers, SREs, DevOps and FinOps teams deliver performant and cost-savvy applications to their users easily and quickly.

The dilemma of cloud-native optimization complexity

The promise of self-healing, highly efficient cloud-native applications is far from being realized. Why is that?

Having worked with many Kubernetes teams in the last years, we are convinced it is because of two root causes: the technical challenges introduced by Kubernetes and the organizational disconnect between teams and their competing priorities.

Technical challenges of Kubernetes applications optimization

Kubernetes resource management is tricky and hard to grasp. How do you properly configure pod resources and horizontal scaling to guarantee application reliability AND resource efficiency across 100s of microservices? This is a key challenge.

Additionally, application runtimes like the JVM or Node.js V8 manage resources in counterintuitive ways when working within containers. The result is a set of hard problems like out-of-memory kills, inefficient resource usage, and performance slowdowns.

Disconnected priorities across teams

On top of the above, there are also organizational issues that make optimization initiatives stall, and the huge potential for both cost savings and performance improvements remains untapped.

Why does it happen?

Platform teams manage Kubernetes clusters but typically can’t optimize applications – that’s the domain of development teams.

Development teams are focused on shipping features and lack both time and expertise for optimization. 

DevOps and SRE are often tasked with optimization activities, but they are overwhelmed with automation tasks and incident handling. 

Even performance engineering teams, despite their expertise, struggle to engage with development teams who lack testing environments and face competing priorities.

There must be a better way.

Introducing Akamas Insights

Today we’re super excited to unveil Akamas Insights for Kubernetes, a new capability that makes it easy for developers, SREs, DevOps and FinOps teams to achieve rapid cost savings, improved reliability, better compliance and, ultimately, higher performance and efficiency in Kubernetes environment, with no effort.

Akamas Insights analyzes your entire Kubernetes environment and provides:

  • Clear visibility into optimization opportunities across all clusters.
  • Estimated impact of the optimization, e.g. achievable savings.
  • Prioritized, safe recommendations for both infrastructure and application configurations.

All of this comes easy, with no skills and effort required to set up, as there are no agents to be installed.

Akamas Insights brings a lot of new capabilities in the space – let’s unpack them.

Automatic optimization opportunities detection

First of all, Insights identifies Optimization Opportunities to improve your Kubernetes environment in three areas:

  • Efficiency
  • Reliability
  • Compliance

Efficiency opportunities identify ways to reduce resource usage and costs. Insights analyzes your Kubernetes environment looking at three layers: clusters, workloads pods, and application runtimes (e.g. JVM or Node.js V8). Thanks to this deep analysis, Insights is uniquely able to identify the biggest optimization opportunities for cost reduction, including inefficient cluster autoscaling or memory wastage due to applications with overprovisioned heap memory.

It’s not just about cost saving. SREs and Dev teams typically care more about application reliability. Insights helps in detecting reliability risks that need to be fixed before impacting production, such as workload with misaligned JVM heap size and memory limits, or misconfigured HPA settings that prevent your app from managing traffic peaks.

Or perhaps you’re a Platform Engineer and you may want to identify workloads that don’t follow best practices. Insights provides you with preconfigured checks to ensure efficient and reliable Kubernetes apps, like memory limits not equal to memory requests or JVMs running with default or inadequate options.

Insights goes well beyond current optimization tools that focus on pod resources alone. It comes with 50+ preconfigured optimization opportunities across the full Kubernetes stack, curated based on our extensive experience and lessons learned optimizing 1000s of Kubernetes clusters.

Estimated savings across infrastructure and application workloads

For opportunities to be adopted, everybody needs to understand their impact. So we built into Akamas Insights a full impact estimation of the detected optimization opportunities.

In a world where teams are overwhelmed with different priorities, the key to successful optimization initiatives is to clearly show engineers the expected benefit they can achieve.

Insights estimates the resource footprint and achievable savings. Furthermore, Insights shows savings broken down by infrastructure and application workloads, so that you can understand the benefits of optimizing your workloads, the clusters, or both.

Application-awareness and safe recommendations

Once you’ve found valuable optimization opportunities, Insights provides you with recommendations to optimize your applications safely and with no effort.

When we talk with Kubernetes teams, we hear a recurring theme: engineers are not willing to optimize their applications. Why? Because they fear impacting application reliability and performance. That’s understandable: current Kubernetes resource optimization tools are simply too basic and just consider container resources, and not the application running inside.

Insights recommendations go much deeper than what current tools offer, and recommends configuration improvements that consider the full application stack. This includes pod resources, HPA scaling policies, and application runtime heap and garbage collection settings for Java and Node.js applications (JVM and V8 runtimes).

The benefit? Doing so, Insights allows you to optimize your applications efficiency, reliability and performance covering all important aspects, with no effort and skills required.

Akamas Insights for Kubernetes

Get started!

Akamas Insights is very easy to set up! No agent to be installed, no security concerns: Insights collects metrics from Kubernetes observability tools. It needs a read-only account to start collecting data and help you optimize your entire Kubernetes environment.

Wanna try it out? We’re excited to invite you to join the Akamas Insights beta program. Experience firsthand how AI-powered, application-aware optimization can transform your Kubernetes operations.

Together, we can make cloud-native optimization accessible, safe, and effective for everyone.

 
Stefano Doni
Stefano Doni

Co-founder & CTO

See for yourself.

Experience the benefits of Akamas autonomous optimization.

No overselling, no strings attached, no commitments.

© 2024 Akamas S.p.A. All rights reserved. – Via Schiaffino, 11 – 20158 Milan, Italy – P.IVA / VAT: 10584850969