The Next Kubernetes Management Frontier: Automation

Originally posted on thenewstack.

Investing in a GitOps-ready, central control plane will point organizations in the right direction of the next Kubernetes management frontier.

Kubernetes as a core technology has become foundational to modern application architectures and continues to expand its market presence. A recent survey of 500 full-time IT department employees conducted by Portworx by Pure Storage finds 87% expect Kubernetes to play a larger role in their organizations over the next two years, with 79% noting they have already increased usage of Kubernetes clusters over the last two years.

The primary reasons for increased reliance on Kubernetes cited by survey respondents are the need to increase levels of automation (56%), followed by reduced IT costs (53%), the need to deploy applications faster (49%), and digital transformation initiatives spurred by the COVID-19 pandemic (48%), the survey finds.

After initial adoption, many enterprise IT organizations quickly realize that Kubernetes is simultaneously the most powerful yet complex platform ever to be deployed and managed. Now those same enterprises are attempting to manage fleets of Kubernetes clusters that present even more networking and security challenges at levels of unprecedented scale.

While it may feel intuitive to run many workloads in a single Kubernetes cluster for easier management and better resource utilization, we observe an increase in the number of Kubernetes cluster deployments, whether it is a consequence of the development team’s own choices or for performance optimization of workloads running at the edge to be closer to users or to isolate workloads for organizational or legal reasons.

Kubernetes was built by some of the world’s most talented software engineers for large-scale architectures. The issue is its complexity requires skilled software engineers which are a scarce resource across today’s highly competitive workforce. Kubernetes expertise is not only hard to find and retain, but the software engineers that have these skills also command some of the highest salaries in the IT industry.

This exponential growth in deployed Kubernetes clusters coupled with the challenges of attracting and retaining in-house Kubernetes expertise leaves small-to-medium-sized IT organizations in a difficult position to keep up with Kubernetes clusters’ sprawl. The market needs simpler ways to industrialize Kubernetes growth at scale, whether through the help of a central control plane or automation, or both.

The Proliferation of Kubernetes Clusters Demands a Central Control Plane

As the fleet of Kubernetes clusters continues to expand, a central control plane is necessary to ensure that the system’s different components can work together effectively and efficiently. Without a central control plane, it would be difficult to manage and coordinate the different Kubernetes clusters, and to ensure that applications are running smoothly end-to-end. This also makes it easier for DevOps and administrators to have centralized management and control over the clusters. A central control plane also needs to take microservices networking into consideration by:

  • Managing global and local traffic from one place while providing a dashboard overview of distributed environments,
  • Applying settings such as traffic management rules and security policies globally across all clusters in a consistent manner and
  • Providing a centralized Global Server Load-Balancing (GSLB) capability to increase reliability and reduce latency for applications spanning multiple regions in public and private clouds.

A centralized control plan with a simple-to-use web GUI is a convenient way to enable teams to quickly bootstrap projects. Organizations that are just getting started with Kubernetes will find this invaluable while they are still handcrafting individual cluster deployments.

But, as organizations accelerate their adoption and use of Kubernetes in production, manual management of multiple clusters becomes untenable.

Automation Is No Longer a “Nice to Have”

The only way to effectively navigate Kubernetes deployments at scale is to adopt the right automation and management tools.

Organizations deploying Kubernetes must make it accessible to the small army of administrators that populate most IT teams. Most individuals seek full automation and audit-ability through GitOps — a version of DevOps automation — to deploy and manage infrastructure and applications across multiple Kubernetes clusters.

At its core, GitOps promotes the use of declarative infrastructure and application definitions, which describe the desired state of the environment rather than the steps required to achieve it. Non-GitOps approaches and deployment strategies for provisioning clusters and deploying manifests are often fragmented and involve manual intervention, which costs engineers time and elongates the process of scaling.

GitOps solves the problem of managing and deploying infrastructure and applications in a consistent and repeatable way with easier collaboration (with full audit trails), version control and the ability to roll back changes. By leveraging GitOps-compliant tools, application teams take advantage of automating the self-healing, autoscaling and observability of Kubernetes clusters, as well as creating a consistent method for incorporating security and observability standards.

Regardless of the motivation behind the initial rise in adoption, it’s clear Kubernetes is now a permanent fixture in the IT landscape as clusters are increasingly deployed everywhere from the network edge to the cloud and everywhere in between. Investing in a GitOps-ready, central control plane will point organizations in the right direction of the next Kubernetes management frontier.

Source: thenewstack