Subscribe

Mind the K8 skills gap

Kubernetes adoption may be on the rise locally, but making the change is complex and needs to be managed carefully.
Joanne Carew
By Joanne Carew, ITWeb Cape-based contributor.
Johannesburg, 22 Feb 2024
Michael Langeveld, HPE.
Michael Langeveld, HPE.

The primary responsibilities of an orchestra’s conductor include setting the tempo, shaping the sound of the ensemble and interpreting the music. They’re in charge of telling the musicians when to start and how fast to go, they rehearse with the orchestra to ensure that each component works well together and act as an artistic leader, making interpretive decisions around how a piece of music should be played to create a memorable experience for the audience.

Kubernetes (K8) acts a lot like a conductor, says Bruce Busansky, application platform specialist at Red Hat Sub-Saharan Africa. And the containers, he adds, are all the musicians. “If you have one without the other, things don’t work as well. A conductor without an orchestra is pretty useless and if we have an orchestra with no conductor, we’re probably not going to hear very beautiful music.” In much the same way, when containers and Kubernetes are brought together, we have a new and modern way of thinking about building, deploying and running business applications, he says.

Containers make it possible for software to run quickly and reliably when moved from one computing environment to another – be it from a physical machine in a datacentre, to a virtual machine in a private or public cloud, or from a developer’s laptop to a test environment. While container-style process isolation has been around since the 1970s, the market took a big leap in the early 2000s when container technology development and refinement took centre stage.

A standard unit of software that packages up code and all its dependencies, containers are a good way to bundle and run applications. This solution has become popular because it provides a number of benefits, from agile application creation and deployment and greater observability to environmental consistency across development, testing, and production and distributed, elastic micro-services.

Containers are also much smaller than virtual machines because they share the OS’ kernel, and can be spun up and down in seconds, versus minutes for a VM.

At best, says Michael Langeveld, head of technology and business development for the Emirates and Africa at HPE, container-based workloads need a lot of automation, orchestration and management. Kubernetes does this. And yet, there is relatively slow adoption of this technology in South Africa. To better understand why this is the case, Langeveld suggests looking at the early adopters. These are typically the companies born in the cloud, which don’t have monolithic legacy applications. They lend themselves well to this approach because they can just get stuck in and develop things from scratch.

Skills, skills, skills

In line with this, Busansky says that it’s not only the newest and more forward-thinking players who are trying this stuff out. There are also those businesses that play around, burn their fingers and are willing to try bleeding-edge stuff because if they fail, they learn something and then come back and do it better, he says. These businesses are oftentimes the banks, some of the big insurance players and the big telcos. So, what’s holding the laggards back?

A MATTER OF SECURITY

According to Red Hat’s “State of Kubernetes Security Report 2023”, organisations are adopting cloud-native technologies like Kubernetes to transform how they build, run

and scale applications. But introducing new technologies can create unforeseen security challenges. In conversations with around 600 DevOps, engineering and security professionals, Red Hat found that:

67% of respondents have delayed or slowed down deployments due to Kubernetes security concerns.

37% have experienced revenue and/or customer loss due to a con-tainer/Kubernetes-related security incident.

38% cite security as a top concern when it comes to container and Kuber-netes strategies.

35% said their current container and Ku-bernetes security solution slows down development.

30% identified vulnerabilities as their biggest worry around their container and Kubernetes environment.

According to a September 2022 Evaluator Group survey of K8 adoption, the shortage of K8 and container management skills is a major challenge. More than half of the surveyed companies that were looking to modernise their application and operations infrastructure struggled to maintain the expertise to adopt this new technology with their current staff compliment. And 35% were having difficulties securing talent.

Having built hundreds of K8 clusters for dozens of large enterprises across the globe, the single most important measure of success of any K8 cluster is support, says Deon Stroebel, chief commercial officer at LSD Open. But this is often hard to find. While the local market for K8 might be growing, there are limited skills available to offer support. “The skills gap is affecting everyone,” he says.

One way to work around a skills gap is to upskill your existing team, rather than trying to hire new people. When asked who is best suited to these upskilling efforts, Langeveld says that businesses across both the public and private sector generally have more people sitting in the infrastructure space than they do in the application and application development areas. “So, it makes sense to reskill and upskill some of these people. But herein lies another challenge,” he says, noting that infrastructure people are typically not all that receptive to change. As such, he suggests that making any changes to your internal skills must be part of a deliberate organisational strategy so that everybody understands what you’re doing and why you’re doing it. This will, hopefully, make them more willing to go along on the journey.

“Typically, infrastructure engineers aren’t skilled coders and K8 is exactly that – it’s infrastructure as code. So, it isn’t surprising that finding engineers who can code is difficult or that this reality can prevent businesses from being able to build a robust enough team to become Kubernetes- proficient,” says Kevin Wilson, GM for IT services at Stefanutti Stocks Corporate Services.

A daunting task

Similarly, adds Wilson, upgrading the skills of the virtualisation engineers of the previous decade so that they can become the container engineers of the next decade is a daunting task. This is because the virtualisation and virtual machine work these engineers are doing continues, even as more and more businesses migrate to new platforms. All of this being said, K8 is definitely the technology of the future, he says, which means that businesses need to come up with ways to access the knowledge and knowhow to ensure that their efforts in this space are successful.

Source: 2022 Cloud Native Computing Foundation (CNCF) Annual Survey.
Source: 2022 Cloud Native Computing Foundation (CNCF) Annual Survey.

Where companies aren’t willing and/ or able to build and develop these skills internally, Busansky says they need to look at how to attract the right skills. “You can put ads on LinkedIn, you can work with an agency, you can ask your friends or your uncle or whoever if they might know someone but, at the end of the day, your company needs to be appealing to young developers to attract the top talent,” he says. “The companies that are managing to do this well don’t have skills shortages because they’re making themselves appear a lot sexier to potential hires. Basically, these companies are inviting everyone to their party and people actually want to come.”

This is a little harder if you’re the Department of Home Affairs, for example, because this isn’t exactly a very sexy place to work with technology. Businesses like this have no option but to outsource some of their technical needs, he says. Outsourcing is also appealing when the tech isn’t what you actually do. “Absa’s business is all about running a bank. It’s not about running K8. King Price insurance does insurance, it doesn’t do K8,” he says. For these businesses, having a competent and reliable partner that will build everything out for them, that will manage and upgrade everything for them and keep all the lights on, just makes good business sense.

Typically, infrastructure engineers aren’t skilled coders and Kubernetes is exactly that – it’s infrastructure as code.

Kevin Wilson, Stefanutti Stocks

“Sure, you could go off and just download community K8 off the internet – anybody can go and do that – but the minute you want to upgrade something or make changes, there’s a high chance that something is going to break,” he says.

For Busansky, any business that is serious about running money-generating business applications on Kubernetes should talk to an enterprise vendor that has the experience and expertise to provide support. “The only thing that matters is that the application is running and that the customer has a positive experience using that application. If you’re logging into Gmail or a Teams call, do you care if it’s running in a container if it’s doing what it’s supposed to do? So it’s less about what’s under the hood and more about what the business can deliver when it has the right technologies in place.”

K8 IN ACTION

According to Forrester Research, examples of Kubernetes-driven innovation are everywhere, with countless brands using this open-source container orchestration system to handle complicated orchestration, to meet evolving business needs, to build new models/products quickly and to enable Agile-plus-DevOps practices. While Kubernetes typically just sits quietly underneath the services we use every day, there are a number of real-world examples and proven use cases that showcase what it can do. Forrester highlights the following:

Audi: Understanding that innovation can be slow if you build every new idea from the ground up, Audi is using managed Red Hat OpenShift services on Amazon Web Services (AWS) and Azure to deliver a customised cloud-agnostic K8 developer environment, called Kubika-O.

CERN: CERN labs use Kubernetes to help process petabytes of data from subatomic particle research. Kubernetes is able to handle spikes in workloads and enables automation, like reducing the amount of time required to add new nodes to complex distributed storage from three hours to less than 15 minutes.

SAS: SAS has long had a presence in analytics and AI. To transform its offerings for the cloud-native era, the company launched the SAS Viya analytics offering on Kubernetes platforms from AWS, Azure and Google.

Mercedes-Benz: An early Kubernetes adopter, Mercedes-Benz built its own solution and shared the results with the open source community. By 2021, Mercedes-Benz ranked among the top 20 companies contributing to Kubernetes and it continues to play a key role in the Cluster API subproject, which looks to simplify the provision-ing, upgrading, and operating of multicluster Kubernetes environments.

Bloomberg: As a provider of financial information, Bloomberg supports traders in the decentralised over-the-counter market. Bloomberg leveraged Kubernetes to build natural language processing to identify offers to sell financial instruments within large group chats. Kubernetes transformed the model development lifecycle for AI by providing the compute needed to power model building with fragmentary and rapidly changing information.

Share

* Article first published on brainstorm.itweb.co.za