Why should you consider Kubernetes? Or why (when) not?

As businesses en masse embrace digital transformation, what’s clear is the need to bolster IT capabilities in a way that supports agile development and rapid innovation. In particular, many organisations are turning to a relatively recent and very powerful addition to digital transformation strategies: containerisation.

Containers package up an application’s code and all of its dependencies to ensure the application runs reliably irrespective of its underlying environment. As such, the containerised application is essentially mobile and able to run nearly anywhere – from a developer's laptop to a test environment, from a test environment to production, or from an on-prem environment to a private or public cloud – without needing to make any changes.

However, as containers proliferate, their overall strength can quickly turn into their Achilles’ heel: How does one manage and orchestrate thousands or even tens of thousands of these dynamic entities with transient lifetimes in large-scale containerised environments? This is where Kubernetes steps in. It brings order to the chaos by making it possible to automatically set up multiple containers that work together and have the ability to scale.

There seems to be a lot of hype around Kubernetes, and it’s fast becoming the de facto container orchestration system for many enterprise IT shops. In this blog, we’ll try to look beyond the hype and help you answer the question: Why should you consider Kubernetes and when it might not really be a good idea?

It handles container orchestration really well!

When you need to run containers at scale, housekeeping can become a hugely complex task. You need to deal with swarms of individual container instances, each performing individual tasks. You need to also find ways of identifying those individual instances, of communicating with them, and of removing them when they become redundant.

Kubernetes eases the burden by efficiently automating the configuration, deployment, management, and monitoring of containerised applications even in the largest-scale environments. 

You can use the platform to handle the scheduling and coordination of containers across clusters, scale them up and down, and efficiently manage the workloads to ensure they run reliably. Kubernetes also includes built-in features for load balancing, which involves distribution of high traffic volumes across numerous container instances.

In a way Kubernetes is more than just container orchestration, however. The platform allows for automatic restart of failed containers, disposal of old ones, and release of updates with near-zero downtime. It provides a simple and easy interface to set and change the desired state for container deployment. You can therefore automatically create new instances of containers in a preferred state, and move existing containers to the new instances, while removing the unnecessary ones.

It increases cloud flexibility

What’s great about Kubernetes is its high degree of portability, which means you can use it on a wide variety of different infrastructure and environment configurations. Most other container orchestrators lack this level of portability since they remain bound to specific container runtimes—the program that actually runs containers—or infrastructures.

This allows you to seamlessly migrate on-premise workloads into the cloud and across multiple clouds, and avoid potential hazards with “vendor lock-in”. In addition, Kubernetes also delivers the ability to scale the workloads across disparate environments. 

The ease with which Kubernetes supports hybrid and multi-cloud strategies has led many major infrastructure vendors to launch Kubernetes-based hybrid cloud offerings. These platforms are designed to not only manage clusters running on-premise and in their own cloud environments, but also clusters deployed in other cloud platforms.

It improves developer productivity

Kubernetes deployment is based on a declarative approach. As mentioned earlier, this gives teams the ability to specify the description of the desired state of resources and supports quick deployments or roll-backs if necessary. 

It’s no surprise that most developers love the platform. With its declarative construct and ops-friendly approach, Kubernetes allows devs to deploy and scale at a faster pace than was ever possible in the past. Instead of spending time in maintenance mode taking care of infrastructure issues, devs can now focus more on development, which is what they’re really out to do. 

It helps you optimise IT costs

Another key benefit is a significant reduction in infrastructure costs in large-scale containerised ecosystems.

Over-provisioning infrastructure made sense in the past because administrators often had the tendency to conservatively handle unanticipated spikes, or just because they found it difficult to manually scale containerised applications.

But, orchestrators like Kubernetes have built-in features such as auto-scaling that enable you to automatically respond based on the needs of your application and the incoming traffic and load processed by your applications. Overall, this leads to greater efficiency in responding to changes in environmental demands and prevents you from paying for resources that you do not need.

It has a large community and adoption

The popularity of Kubernetes has given it a broad community of end-users, contributors, and maintainers, whom you can rely on for support and advice when you’re faced with technical issues. 

Plus, there is now a rich ecosystem of add-ons and complementary software applications, which extend the functionalities and range of capabilities offered by the platform. For instance, if you have a specific requirement that Kubernetes cannot meet adequately, there is a reasonably good chance that there exists an add-on to address your particular use case. 

It's the de facto standard for deploying containers in production

Because containerisation technology has been around for some time now, initially there were quite a few different open-source projects competing to become industry standards. However, we've now come to a point where the industry has generally standardised on Kubernetes. 

Thousands of IT teams are using Kubernetes on a daily basis in a lot of different ways, thus proving it is reliable, stable, and battle-tested. It’s also worth noting that almost all major cloud computing platforms and providers like AWS, Google Cloud, Microsoft Azure, etc. now offer support for Kubernetes.

Now that we have explored the benefits associated with Kubernetes, we can come to the question of when you shouldn’t use it or what factors you should keep in mind before you switch to Kubernetes.

If your application doesn’t use a microservices architecture

It doesn’t make sense to use Kubernetes if your application architecture doesn’t follow a microservices approach. In the case of a conventional monolithic architecture, it’s not always best practice to use containers and a tool to orchestrate them, although it can be achieved. 

While every piece of the application—from IO to data processing to rendering–is intertwined in a monolithic architecture, containerisation involves breaking your application down into individual components (microservices). Kubernetes is in alignment with a typical microservices architecture, where you have several individual components that work together and might need some complex initialisation and setup.

Therefore, companies which operate a complex microservices environment are more likely to see the real benefits of container orchestration tools like Kubernetes.

If your team doesn’t have adequate skills

Kubernetes is a complex technology with a lot of moving parts and a notoriously steep learning curve. Learning, setting up, and utilising the platform is a specialisation on its own. 

You need to spend a good amount of time and resources to educate your DevOps teams and ensure that they’re ready to operate it. Adequate experience, continuous practice, and extensive training are critical, especially for developers not familiar with infrastructure automation technologies, so they become familiar enough to be able to debug and troubleshoot.

If you’re not prepared to take transitional challenges into account

Even with skilled DevOps staff on your side, the transition to Kubernetes might still be cumbersome and require a large investment of effort and time.

Because most companies cannot start with a greenfield project, you need to find ways to make sure your existing software can run smoothly alongside Kubernetes. 

It can sometimes be difficult to specify the precise amount of effort you may need to adapt your existing software, since this depends on the kind of software being used (for instance, whether it is already containerised, which programming language is used, etc.). 

Plus, you need to adapt your existing processes, especially deployment processes, so they can work optimally in the new environment.