Azure Kubernetes Services

In this article we will see about Azure Kubernetes Services and some of its case studies.

So let’s begin….

The use of containers to develop and deploy software has become popular over the last few years. Containers make it easy to package and deploy an application with all its services to any compute environment. When your application meets higher demand, you can easily scale out your services by deploying additional container instances.

Scaling multiple containers becomes challenging as several factors need consideration when managing multiple containers. Suppose you need to handle load balancing, security, network connectivity, and deployment. To help make this process easier, it’s common to use a container management platform such as Kubernetes.

AKS:

Azure Kubernetes Service (AKS) manages your hosted Kubernetes environment and makes it simple to deploy and manage containerized applications in Azure. Your AKS environment is enabled with features such as automated updates, self-healing, and easy scaling. The Kubernetes cluster master is managed by Azure and is free. You manage the agent nodes in the cluster and only pay for the VMs on which your nodes run.

How workloads are developed and deployed to AKS

AKS supports the Docker image format that means that you can use any development environment to create a workload, package the workload as a container and deploy the container as a Kubernetes pod.

AKS also supports all the popular development and management tools such as Helm, Draft, Kubernetes extension for Visual Studio Code and Visual Studio Kubernetes Tools.

Case Study:

  1. Hafslund Nett

Hafslund Nett (Hafslund), the power grid operator that serves 1.5 million Norwegians, determined that legacy systems for reading meter data needed higher capacity and that externally developed software was difficult to manage.

Based on experience with large-scale projects in other corporate divisions, Hafslund recognized the value of containerized applications and the efficiencies of scale that they offer. Hafslund chose to develop its own meter-system software, using Microsoft Azure as its cloud platform, Azure Kubernetes Service (AKS) to manage software containers, and Azure Monitor for containers to optimize container performance.

They wanted highly secure networking between on-premises IT resources and Azure, which was a prerequisite for workloads they wanted to run with Kubernetes. Additionally, Azure Active Directory readily supports role-based access control on Kubernetes clusters, which is a security measure that aligns with the company’s critical best practices.

The company is building a platform that uses AKS to support not only meter-reading but also most internally developed software covering three broad areas:

  • Data integration — that is, microservices to implement representational state transfer (REST) interfaces used by an internal central data integration tool
  • APIs that expose data both internally and externally
  • Complex systems consisting of several interoperating services that interact directly with users

2. Bosh:

Robert Bosch GmbH set out to solve the problem of drivers going the wrong way on highways, the goal was to save lives. Other services like this existed in Germany, but precision and speed cannot be compromised.

When the product team brainstormed the idea to solve the problem of wrong-way driving, they did not know whether it was technically possible. For such a service to work commercially, it had to locate vehicles in real time with pinpoint precision.

Smartphones or an onboard connectivity unit can anonymously record GPS coordinates and can send that location data to the cloud if the device is in a hotspot area, but GPS satellites broadcast their signals in space with only limited accuracy.

They had to solve two major issues: first, to get the last piece of information out of the noisy sensor data; and second, to develop a highly scalable and ultra-flexible service to process the data in near real time.

The problem was speed. The team assumed that devices emitting location information, such as smartphone apps and automotive head units, could eventually send thousands of data points to the solution per second, from all over Europe and eventually other countries. Bosch needed lightning fast compute capable of filtering events and pushing a notification back to an end device within 10 seconds — the time estimated to make the solution viable.

The key was orchestration. By orchestrating the deployment of containers using AKS, Bosch would get repeatable, manageable clusters of containers. Bosch already had a continuous integration (CI) and continuous deployment (CD) process to use in producing the container images and orchestration. The result: increased speed and reliability of deployments.

Architectural overview of the Bosch wrong-way warning driver service on Azure

By running their solution on Azure and AKS, the average time to calculate whether a driver is going the wrong way could be improved to approximately 60 milliseconds.