How To Provision a Production-Grade Kubernetes Cluster From Anywhere

Remember?

I bet everyone who're working or having fun with Kubernetes still remember perfectly the very first time you attempted to set up it.

And so the second time.

And so the third time.

…

And lastly, the one which it labored.

And many likely, if you are an expert additionally you recall the lengthy path that introduced you to definitely own the needed expertise on Kubernetes to be able to setup and fine-tune production-grade clusters to operate apps.

Or, if Kubernetes isn't a part of your job’s scope, you most likely remember the length of time it required that you should hire a company capable of singing a legitimate Kubernetes install…and just how much it costed.

In order to save all of this effort and time to the customers Cisco released Cisco Container Platform (CCP), a turnkey means to fix easily provision production-grade Kubernetes clusters on-prem or perhaps in the cloud within a few minutes, with couple of clicks and requiring virtually no understanding of K8s. All of the needed integrations when it comes to network, storage and security are carried out instantly by CCP so the provisioned K8s clusters will be ready to run being produced.  Clusters provisioned by CCP happen to be outfitted with correctly-configured monitoring and logging tools like ElasticSearch, FluentD, Kibana. With the Container Network Interface (CNI) you may choose whether or not to leverage Cisco ACI as network infrastructure or any other ones for example Contiv or Calico (no reliance on the actual infrastructure). With CCP you are able to take proper care of the entire existence-cycle from the K8s cluster: it is simple to perform Kubernetes software upgrades, nodes upgrade, cluster scale up or lower and cluster deletion.

This really is already good so if you're following our cloud bulletins you may know this, and so i thought I’d produce a demo that could push the simplicity individuals “few mouse clicks” to the limit, making possible to produce a production-grade cluster in only a single click - literally.

Presenting the Kubernetes dash button.

The idea is rather simple: develop a dash button that, once pressed, results in a production-grade Kubernetes cluster available.

Leveraging the wealthy group of the Cisco Container Platform (CCP) APIs this really is even too easy, and so i considered to then add more feature on the top:

• I needed to provision the cluster and can get on just with the dash button. So, I needed CCP to show around the dash button itself the Ip from the master node from the cluster produced
• I needed bi-directional communications between your dash-button and CCP itself, so will be able to check up on the dash button if CCP properly received the provisioning request, and make certain the provisioning process has began after which finished.
• I needed a good battery existence that will avoid me getting to recharge the button every single day, and so i required to have electronics in a position to sleep or hibernate
• My lab, where I've the infrastructure and CCP, is behind a proxy, and for that reason not accessible in the outdoors world, which meant I'd to find away out to possess my lab initiate communication using the dash button by positively examining the press from the button
• I needed to make use of the button everywhere I am going without having to worry concerning the local Wi-Fi settings

How it operates

To fulfill all of the above needs I added a few elements within the picture, winding up using the following architecture:

The button is dependant on an Arduino ESP 32 board, it connects via Wi-Fi to my smartphone and uses its web connection, by doing this I'm able to make use of the button everywhere my phone has data signal. A publish-subscribe message service (MQTT) on the web is accustomed to bypass the proxy limitations. I located the MQTT in your own home however, you can provision one on AWS or make use of a free MQTT service around the cloud. Once pressed, the button publishes a unique message around the MQTT service. Within my lab, a few scripts are continually polling the MQTT service and, every time they identify the special message, they invoke the best API within the Cisco Container Platform to trigger the provisioning of the shiny new Kubernetes cluster. When the cluster is provisioned, the Ip from the master node is came back, with the MQTT service, towards the dash button that shows it on its display, and, at this time the Kubernetes cluster is able to accept connections and run applications.