Kubernetes

is an open-source container orchestration tool, over a period of time its usage has increased rapidly, and many more companies began to adopt Kubernetes, which created a requirement to get Kubernetes as a managed service, so that companies can focus more on their applications, that’s where google Kubernetes engine comes in.

These are some of the components of a Kubernetes cluster you will have to understand in order to run/create a open-source Kubernetes cluster from scratch.

• Control plane provisioning and management
• Worker nodes provisioning and management.
• Security and network configuration
• Patching and upgrade
• Monitoring
• Scaling
• Availability and Reliability

 

What is Google Kubernetes Engine?

Google Kubernetes Engine is a managed Kubernetes service provided by google cloud platform, it allows you to create a Kubernetes cluster quickly, with many of the above-mentioned complex components deployed and managed by GCP, this allows you to focus more on developing and running the application on the managed Kubernetes cluster, rather than spending a lot of time in understanding and managing Kubernetes cluster itself.

Google Kubernetes Engine provides you two operation modes Standard mode and Autopilot mode, both autopilot mode and standard mode GKE clusters take some of the management overhead of Kubernetes clusters, each of them will also allow different level of interaction with Kubernetes cluster.

Standard mode in GKE

Standard mode will create control plane, deploy worker nodes and security, network configuration is also handled by GCP, but you need to manage the worker nodes of the Kubernetes cluster, like configuring autoscaling and managing upgrades, it provides you with all four dimensions of auto scaling, but you will have to manually configure auto scaling, here you will pay per node and the instance types that you select.

Auto Pilot mode in GKE

It provides an even more hands free experience than standard mode, with smart defaults and requiring less management overhead GKE auto pilot mode allows you to concentrate more on your own application, as the nodes of the cluster are automatically provisioned and managed, there is no need to dedicate time and resources to node management and upgrades and all the infrastructure autoscaling will also be handled automatically by the auto pilot mode itself, here you will pay per pod or per-workload, rather than per-node

if you want minimal cluster-management over head in order to focus more of your time and resources on the application you need to run, you may prefer GKE auto pilot mode.

Now let’s go through the creation of Standard and Auto pilot mode GKE clusters, so that you will get a complete understand of both

 

Configuring GKE Autopilot mode cluster

on your GCP web console go to Kubernetes dashboard and select to create Autopilot cluster, then select a Region to deploy your clusters control plane and worker nodes. Autopilot Clusters are only available on regional level.

And for network access choose Public Cluster if you want to allow access to your cluster from public networks, choose Private Cluster to block access from public networks and allow communication only via private google cloud network, you can configure cloud network address translation for your nodes allowing them to send outbound requests to the internet

Note: you cannot change the cluster type once it is created.

Next in Advanced options you can configure release channels, maintenance period and give description and labels to your GKE cluster

Release channels

select a Release channel for your GKE cluster, google automatically manages the version and upgrades for the cluster and its node pools based on the release channel you select, your choice of release channel between Rapid and Regular and Stable will affect how quickly new versions of Kubernetes become available for your cluster to upgrade and how often the cluster should upgrade to stay in compliance with the currently supported versions.

Automation

Select for Enable Maintenance window under Automation to give a specific maintenance window for your cluster normally maintenance can happen at any time if not configured, you must allow at least 48 hours of maintenance availability in a 32-day window, and a maintenance window must be at least of four hours.

select security options based on your requirement and giving labels will help you in organizing your GCP resources.

 

 

Configuring GKE Standard mode cluster

Now let’s look at the configuration of a GKE Standard mode cluster, on your Kubernetes dashboard of GCP web console click on CREATE and select standard mode, a brief section wise explanation is given below.

 

as usually give a Name to your cluster and then select a zone or a region.

• As mentioned, Autopilot mode clusters can only be regional but in standard mode you can configure regional cluster also zonal cluster within a region.

And then select a Release channel to configure frequency of Kubernetes upgrades, a brief explanation about release channels is already given in autopilot mode configuration steps, however in standard mode you can also create a GKE cluster using a static version of Kubernetes, which will increase your responsibility in managing your clusters version periodically.

 

Node pools

here you can configure the number of worker nodes for your GKE cluster, you can also see information about the nodes and select the machine type, and the image of operating system that will be configured on those nodes.

In security you can configure service accounts, cloud API access scopes,

Next in the Metadata section, you can give labels specifically to the nodes of your cluster, rather than giving labels to the whole cluster, here you can also give Kubernetes node taints which will help you to manage what workloads will be able to run on a specific node pool.

 

Next in the cluster section you can configure automation, networking, security, meta data and additional cluster features for your cluster, in Automation section you can configure maintenance windows, and in Networking section, you can configure public or private cluster, as we saw in autopilot mode.

 

Please contact our team for any offshore infrastructure management services website, LinkedIn