Now Pay-per-Pod for your Kubernetes Application!
akhilsharmaGCPkubernetes

Now Pay-per-Pod for your Kubernetes Application!

Google cloud platform (GCP) has recently come up with a new feature for their Kubernetes cluster. With this feature, you don’t need to pay for the whole Kubernetes cluster but only pay for pods you spin up! This new feature is called Autopilot mode!

With Autopilot, you no longer have to anticipate the compute capacity of your workloads or monitor the health of your nodes. Google does that for you!

Here’s what Google says about their new product:
Autopilot is a new mode of operation in Google Kubernetes Engine (GKE) that is designed to reduce the operational cost of managing clusters, optimize your clusters for production, and yield higher workload availability.

For those new to Google Cloud, Google does offer a standard mode for GKE cluster where you manage the cluster’s underlying infrastructure, giving you node configuration flexibility. It’s just that we now have a new mode that’s Autopilot! Here, GKE provisions and manages the cluster’s underlying infrastructure, including nodes and node pools, giving you an optimized cluster with a hands-off experience.

The best part of using Autopilot is that this model works on pay-per-pod, so you are only pay for the resources you use. You don’t need to pay for the nodes or any extra unutilized resources. Also Autopilot comes with an amazing SLA which means you are well protected from any kind of downtimes as well. And incase SLA is breached, you do get credits for that!

SLA in Autopilot Mode

You must be wonder that how does Autopilot know how much resources to configure? Well for this, Autopilot uses the resource requirements that you define in your PodSpec and provisions the resources for the deployment such as CPU, memory, and persistent disks. Just incase you need to scale your applications, all you need is to configure Horizontal Pod Scaling or Vertical Pod Scaling and all your pods would be scaled as per your needs. So, Kubernetes cluster management comes to almost zero when working with Autopilot mode instead of Standard mode. You may read more differences between the two here: https://cloud.google.com/kubernetes-engine/docs/concepts/autopilot-overview#comparison

Autopilot is awesome and makes life really simple when it comes to deploying your Kubernetes applications. But before you start planning to move your next application to Autopilot, consider the following limits:

  • HostPort and hostNetwork are not permitted/accessible because node management is handled by GKE.
  • Not all values of OS and arch are supported in Autopilot.
  • Autopilot does not support Container Threat Detection.
  • Privileged mode for containers in workloads is mainly used to make changes to nodes, like changing kubelet or networking settings.
  • You cannot create certificate signing requests within Autopilot.

Check the complete list of limitations and restrictions on this link: https://cloud.google.com/kubernetes-engine/docs/concepts/autopilot-overview#limits

I hope you find this helpful and this makes your Kubernetes journey a step easier.