Insights for Kubernetes

What is Insights

Insights is a new Akamas capability that helps SREs, platform engineers, developers and FinOps teams uncover hidden cost inefficiencies and reliability risks in your Kubernetes clusters and applications.

Insights provides actionable recommendations to optimize your Kubernetes environment quickly and easily, without requiring setup effort and skills.

Why Insights

Achieving reliable and cost-efficient Kubernetes clusters and applications is easier said than done. The untold reality is that most Kubernetes clusters are massively over-provisioned, and at the same time, applications suffer reliability issues.

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. For more information, read our launch blog.

Why Insights is different

  • No agents required: no setup time, no security checks are required.

  • Full-stack optimization approach: while current Kubernetes optimization tools just consider pod CPU and memory resources, Insights goes deeper inside the pod and optimizes the application runtime, such as the JVM for Java applications or V8 for Node.js applications. This is unique in the industry.

  • No effort required: identifies optimization opportunities and provide recommendations with no effort and deep Kubernetes and application runtime skills required.

  • Designed with safety in mind: recommendations are full-stack and consider the application running within the pod. This avoids reliability risks such as out-of-memory errors or CPU throttling, hence are trusted by development teams.

  • Best practices built-in: provides not only recommendations but also best-practices your teams can use to avoid reliability incidents and run highly efficient Kubernetes environments.

How Insights works

  1. Connect Insights with your Kubernetes observability solution Insights collect metrics from your existing observability tools. See the FAQ for the list of supported tools.

  2. Insights gathers metrics history of your Kubernetes clusters See below for more details about which data is collected.

  3. Insights analyzes collected data using its full-stack, application-aware recommendation engines and knowledge base Insights analyzes data and identifies opportunities to optimize efficiency and reliability using its full-stack, technology-specific recommendation engines. Recommendations are generated considering clusters, workload and application runtimes like the JVM.

  4. Insights shows the identified cost savings opportunities and reliability issues, plus recommendations to improve Kubernetes efficiency and reliability

Example screenshot

Insights summary dashboard showing optimization opportunities across all clusters, and a recommendation to optimize the pod resources and JVM memory for a Java application.

Integration requirements

Insights collects data leveraging the observability tool you are already using to monitor your Kubernetes environment. No agent needs to be installed on your clusters.

Account credentials

Insights simply needs a read-only account to connect and extract data from your observability tool.

Type of collected data

Insights collect technical metrics and configuration information only (see below for details). No PII information is collected.

Metrics collected

Insights analyzes and provide recommendations to optimize the full Kubernetes stack.

To do so, it requires access to the following metrics:

Level
Description
Examples

Kubernetes cluster

Metrics and configuration information related to

  • cluster

  • nodes

  • cluster autoscalers

  • Cluster CPU/memory requests, limits, and used

  • Node CPU/memory requests, limits, and used

Kubernetes workloads

Metrics and configuration information related to

  • workloads

  • pods & containers

  • HPA

  • namespaces

  • resource quotas

  • Pods CPU/memory requests, limits, and used

  • HPA replica count

  • Namespaces CPU/memory requests, limits, and used

Application runtime

Metrics and configuration information related to the runtime powering the application

  • Java virtual machine (JVM)

  • Node.js V8 (planned)

  • JVM heap size, usage

  • JVM garbage collection

  • JVM configuration

Not all metrics are mandatory!

We recommend to feed Insights with all the mentioned layers for best results. However, not all the layers are mandatory. In particular, application runtime metrics are used by Insights to optimize your applications for max reliability and efficiency. However, if application runtime metrics are not available in your observability tool, Insights will still provide technology-agnostic recommendations.

Getting started

Insights is in beta status and will be in GA soon. Try it out and give us your feedback!

Request your access here.

Frequently Asked Questions

Do I need to install anything in my cluster? No — Akamas Insights is agentless. It leverages metrics already collected by your Kubernetes observability tool. Which observability tools are supported? Observability tools currently supported are:

  • Dynatrace SaaS

  • Datadog

  • Prometheus (planned)

  • Grafana Cloud (planned)

We're adding support for more solutions, please reach out to us if your solution is not listed here. What is the deployment model? Insights is a SaaS-based solution. Will this modify workloads? No — Insights is read-only and does not modify your workloads. You can inspect the recommendations and apply them manually. Support for automation is planned.

Can I use Insights with multiple clusters? Yes — Insights supports multi-cluster views and analysis.

Last updated

Was this helpful?