Storage Quality in Kubernetes: How to Optimize Your Resources

Kubernetes v1.22, released in August 2021, introduces a new alpha feature that enhances how Linux nodes manage memory resources. This advancement marks a significant step forward for developers and DevOps teams in resource management.

What Specifically Changes for Developers/DevOps Teams?

In previous versions, Kubernetes did not support quality guarantees for memory. This meant there could be issues when allocating memory resources, especially if containers exceeded their limits. With the introduction of Memory QoS (Quality of Service), you can now ensure that critical processes are always allocated sufficient memory.

An example of a Pod configuration might look like this:

apiVersion: v1
kind: Pod
metadata:
  name: example
spec:
  containers:
  - name: nginx
    resources:
      requests:
        memory: "64Mi"
        cpu: "250m"
      limits:
        memory: "64Mi"
        cpu: "500m"

Here, spec.containers[].resources.requests and spec.containers[].resources.limits are crucial for planning and executing your containers. While CPU is considered a “compressible” resource, this was not the case for memory in cgroup v1. If a container exceeded its memory limit, it would be terminated by the kernel with an OOM (Out of Memory) kill.

Thanks to the new cgroup v2, Kubernetes now offers an improved implementation that ensures complete protection for memory resources. Memory QoS uses the memory controller of cgroup v2 to ensure that your Pods always have the necessary memory resources.

Practical Examples or Use Cases

When you set memory.min to the memory requests, this memory is reserved and never reclaimed by the kernel. This guarantees the availability of memory for your Kubernetes Pods. Additionally, memory.high throttles memory usage when containers reach their limits to prevent system overload.

Here is a table detailing the functions of the two parameters and their correspondences to the resources of Kubernetes containers:

   </td>

With these new features in Kubernetes, you can not only increase the reliability of your applications but also improve the efficiency of your resource management. ayedo is proud to be a partner in this innovative area of Kubernetes development and is happy to assist you in implementing these new capabilities.

Source: Kubernetes Blog