Nov 28, Paris: AI Health Hack #2 with Alan, Qovery and AWS. Applications close Nov 21
Topic · Kubernetes

Kubernetes Operations

Kubernetes operations, often called day-2 operations, is the ongoing work of keeping production clusters upgraded, scaled, secured and cost-efficient after they are first provisioned.

Creating a cluster takes an afternoon. Keeping it current takes a calendar. Upstream Kubernetes ships about three minor releases a year, each managed service retires old versions on its own schedule, and staying on a version past standard support costs extra on EKS and AKS.

Most of the effort sits around the control plane rather than in it: node pools and autoscaling, CNI and ingress choices, add-on versions, the boundaries between teams (namespaces, virtual clusters or separate clusters), and the observability that tells you when something drifts.

These guides cover each of those layers on EKS, AKS and GKE. Qovery provisions and upgrades Kubernetes clusters in your own AWS, GCP, Azure or Scaleway account, or connects to a cluster you already run, so day-2 work does not need its own team.

In this topic
Start here

What is Kubernetes? The reality of Day-2 enterprise fleet orchestration

What Kubernetes is and what it does, why provisioning one cluster is the easy part, and what running a fleet actually demands.

Apr 11, 2026 · 10 min

Day-2 operations: an executive guide to Kubernetes scale

For most enterprises, a Kubernetes migration is pitched as a finite project. In reality, Day-1 deployment is merely the start of an infinite marathon. As clusters scale, a quiet financial crisis emerges. When your highest-paid senior engineers spend 40% of their week on cluster upgrades, certificate rotations, and manual node scaling, you are paying a permanent tax on innovation.

Mar 19, 2026

Day-0, Day-1, and Day-2 operations: the enterprise guide to Kubernetes lifecycles

Day-0, Day-1 and Day-2 operations explained, what each phase actually requires, and why Day-2 is where most Kubernetes cost hides.

Mar 17, 2026

Kubernetes 1.34 Hits End of Life: How to Upgrade EKS, GKE and AKS Without Breaking Production

Kubernetes 1.34 is losing upstream support, and EKS, GKE and AKS each run a different clock. Here is the exact upgrade order that does not break production, the removed APIs to catch first, and what each cloud charges you for running late.

Oct 8, 2026

Kubernetes 1.34 Hit End of Life on EKS: Find Every Affected Cluster Before the 6x Bill Lands

When an EKS cluster passes end of standard support, AWS auto-enrolls it in Extended Support and the control plane jumps from $0.10 to $0.60 per hour - 6x, per cluster. Here is the exact sweep to find every affected cluster across accounts and regions, the pre-flight checks that matter, and the upgrade path.

Oct 29, 2026

AKS LTS Moves to Kubernetes 1.37: How to Get Ahead of Azure's Forced Upgrade

Azure is making Kubernetes 1.37 the next AKS Long Term Support release, and AKS automatically upgrades clusters that fall out of support. Here is what changes, what actually breaks, the real cost of LTS, and a step-by-step plan to upgrade on your schedule instead of Azure's.

Nov 7, 2026

Upgrading Hundreds of Kubernetes Clusters: Interview with Pierre Mavro, CTO of Qovery

n this interview, Pierre explains the process, tooling, and testing strategy in upgrading clusters at scale.

May 15, 2024

Everything you need to know about Kubernetes autoscaling at fleet scale

When engineers configure pod autoscaling, they instinctively tie the Horizontal Pod Autoscaler (HPA) to CPU utilization. If the application is actually bound by memory or downstream database connections, the cluster sits idle while incoming requests time out. Diagnosing hundreds of outages reveals a clear pattern: effective elasticity requires scaling on custom application queues, not just default hardware thresholds.

Mar 30, 2026

Configuring Karpenter at scale: advanced Day-2 node provisioning and Finops

How to configure Karpenter for production: NodePool design, disruption policies and the dual-NodePool pattern, from running it at scale.

Jan 3, 2026

Kubernetes Cluster Autoscaler vs Karpenter

One of the most exciting things when using Kubernetes is the ability to scale up and down the number of nodes based on application consumption. So you don’t have to manually add and remove nodes on demand and let it go on usage. Obviously what you want is to keep control on the minimum and the maximum number of nodes to avoid an unexpected bill.

Jul 5, 2022

Managed Kubernetes comparison: EKS vs. GKE for multi-cloud fleets

When comparing Amazon EKS and Google Kubernetes Engine (GKE), GKE often provides a more automated, hands-off experience with its Autopilot mode and rapid release channels. EKS excels in hybrid cloud integrations and government cloud support. However, at fleet scale, organizations frequently use both, requiring an agentic control plane to enforce global cost governance and standardize Day-2 operations across multi-cloud environments.

Mar 30, 2026

EKS vs. AKS: choosing the right managed Kubernetes for enterprise scale

EKS vs AKS compared on control plane cost, cluster and node limits, identity, upgrades and compliance, with guidance on which to pick.

Feb 14, 2026

GKE vs EKS without a platform team: 7 ways to run Kubernetes in 2026 (ranked)

GKE Autopilot is easier out of the box than EKS, and EKS Auto Mode has closed most of the gap - but neither gives developers a self-service workflow. Here are 7 ranked ways to run Kubernetes on GCP or AWS with no dedicated platform team, with real pricing, cited sources, and honest trade-offs.

Oct 17, 2026

AKS vs EKS for a Small DevOps Team: Which Is Easier to Run, and What You Still Add on Top

AKS vs EKS for a team with one or two platform engineers, checked in November 2026: which one is easier to manage, what the control plane and support tiers really cost, and the layers you still add on top of either one (ingress, certificates, autoscaling, observability, secrets, CI/CD, preview environments, upgrades, cost visibility).

Oct 11, 2026

Kubernetes multi-tenancy in 2026: namespaces vs vCluster vs separate clusters, compared

Namespaces, Capsule, HNC, vCluster, Kamaji, sandboxed runtimes, or one cluster per tenant: what each Kubernetes multi-tenancy model actually isolates, what it costs per tenant, and who should operate it.

Oct 31, 2026

Cilium vs Calico: Which Kubernetes CNI Should a Small Platform Team Pick in 2026?

A decision guide to Kubernetes CNI choices for platform teams with no dedicated networking engineer: where Cilium wins, where Calico wins, when your cloud's managed dataplane beats both, and when the right answer is to stop picking a CNI yourself.

Nov 1, 2026

Kubernetes network isolation and NetworkPolicy at fleet scale

Kubernetes NetworkPolicy acts as a native firewall to control pod-to-pod communication. Because Kubernetes allows all lateral traffic by default, configuring network isolation manually creates massive configuration drift. Agentic control planes automate zero-trust compliance, deploying and reconciling network policies simultaneously across thousands of global clusters.

Feb 7, 2026

Kubernetes multi-cluster architecture: solving day-2 fleet sprawl

Kubernetes multi-cluster management is the Day-2 operational practice of orchestrating applications, security, and configurations across geographically distributed clusters. Because native Kubernetes was designed for single-cluster orchestration, enterprise platform teams must implement a centralized control plane to prevent configuration drift and manage a global fleet without scaling manual toil.

Mar 4, 2026

10 best Kubernetes management tools for enterprise fleets in 2026

Ten Kubernetes management tools compared for enterprise fleets, covering multi-cluster operations, upgrades, cost control and governance.

Apr 20, 2026

Best Tools for Managing Amazon EKS Clusters: 12 I'd Actually Run in 2026

A category-by-category buyer's guide to the best tools for managing AWS EKS clusters in 2026 - eksctl, Terraform, Karpenter, K9s, Lens, Kyverno, Falco, Argo CD and internal developer platforms - with seven comparison tables, cited pricing and CNCF status, and a recommended stack by team size.

Sep 7, 2026

Scaling Kubernetes on AWS: Day-2 operations and fleet management

Scaling Kubernetes on AWS requires balancing node-level elasticity with pod-level demands. A critical architectural oversight occurs when teams misconfigure the Horizontal Pod Autoscaler (HPA) and Vertical Pod Autoscaler (VPA) to compete over the same metrics, causing scaling thrash. Effective Day-2 operations demand standardizing these configurations across your fleet to prevent runaway cloud costs.

Mar 15, 2026

Kubernetes cost optimization: agentic FinOps for enterprise fleets

The three pillars of Kubernetes spend (Compute, Network, and Storage) compound rapidly at enterprise scale. While manual cost-cutting works for a single cluster, managing 1,000+ clusters requires an agentic FinOps approach. By automating resource right-sizing, Spot instance orchestration, and idle environment shutdowns, organizations can eliminate cloud waste without sacrificing production stability.

Feb 7, 2026

Kubernetes observability at scale: how to cut APM costs without losing visibility

The instinct when setting up Kubernetes observability is to instrument everything and send it all to your APM vendor. That works fine at ten nodes. At a hundred, the bill becomes a board-level conversation. The less obvious problem is the fix most teams reach for: aggressive sampling. That is how intermittent failures affecting 1% of requests disappear from your monitoring entirely.

Apr 21, 2026

What Does Kubernetes On-Premise Really Cost in 2026? A Line-by-Line Comparison With EKS, GKE and AKS

A realistic, line-by-line breakdown of what running Kubernetes on your own hardware costs versus a hyperscaler's managed service (EKS, GKE, AKS) - servers, racks, licenses, control plane fees, and the platform engineers you need either way.

Oct 12, 2026

Frequently asked

What are Kubernetes day-2 operations?
Day-2 operations are everything that happens after a cluster is running: minor version upgrades, add-on and node image updates, autoscaling, certificate rotation, security patches, backups, monitoring and cost control. Day 0 is design and day 1 is the initial install.
How often do you need to upgrade Kubernetes?
Plan for at least one minor version upgrade a year. Upstream Kubernetes releases about three minor versions a year and each gets roughly 14 months of patches. On EKS a version gets 14 months of standard support, then 12 months of extended support at a higher cluster fee; AKS and GKE follow similar windows with their own long-term support options.
Should I choose EKS, AKS or GKE?
Usually the cloud where your data, credits and team already are. GKE has the most automated upgrades and Autopilot mode, EKS has the largest ecosystem and Karpenter, and AKS fits teams on Microsoft Entra ID and Azure agreements. The day-2 work is similar on all three.
Karpenter or Cluster Autoscaler?
Cluster Autoscaler resizes predefined node groups and works on every major cloud. Karpenter picks instance types per pending pod and consolidates nodes, which tends to cut cost and scale faster; it is mature on AWS and available on Azure through AKS node auto-provisioning.
Namespaces or separate clusters for multi-tenancy?
Namespaces with RBAC, network policies and resource quotas are enough for trusted internal teams. Virtual clusters add a separate control plane per tenant at low cost. Separate clusters give the strongest isolation for untrusted tenants, regulated data or different upgrade schedules, at the price of more clusters to maintain.

Keep your clusters current without a platform team.

Qovery provisions, upgrades and scales Kubernetes in your own cloud account, and deploys onto clusters you already run. The Free plan covers one cluster of your own.