DevOpsCraftDevOpsCraft
All Services
Container Orchestration

Kubernetes Setup & Consulting

Production-ready Kubernetes from scratch — or a health check on what you have. GitOps, security hardening, autoscaling, service mesh, multi-cluster. Your team ships with confidence on day one.

EKSGKEAKSk3sRKE2OpenShiftArgoCDFluxIstioLinkerdKarpenterKEDAFalcoKyvernoVelero

What we cover

Six domains — every layer of a production Kubernetes environment.

Cluster Provisioning

Right-sized, production-ready cluster from scratch. Managed or self-hosted. Multi-AZ, private networking, node group strategy.

EKSGKEAKSk3sRKE2 (Rancher)OpenShiftTalos LinuxKubeadmKubesprayK0s

GitOps & Deployment

Declarative, git-driven deployments. Every change tracked, every rollback one command. No manual kubectl apply in production.

ArgoCDFlux v2HelmKustomizeHelmfileArgo RolloutsFlaggerWeave GitOps

Networking & Ingress

CNI selection, ingress controllers, TLS automation, DNS management, load balancing strategies, and cross-cluster connectivity.

Nginx IngressTraefikHAProxyEnvoyGateway APIcert-managerExternalDNSMetalLBCiliumCalicoFlannelWeave Net

Security & Policy

Defense-in-depth: pod security, network policies, admission control, runtime threat detection, image signing enforcement.

KyvernoOPA / GatekeeperFalcoFalcosidekickPod Security AdmissionTrivy OperatorCosignKubescapekube-benchPolarisNeuVector

Autoscaling & Resources

Scale workloads and nodes automatically based on traffic, queue depth, or custom metrics. Never over-provision or under-provision.

HPAVPAKEDACluster AutoscalerKarpenterCAST AIGoldilocksLimitRangeResourceQuota

Backup & DR

Automated backup of cluster state, PVs, and namespaces. Tested restore procedures. Cross-region disaster recovery.

VeleroKasten K10LonghornStashetcd backupRook/CephEBS SnapshotsGCS Snapshots

Full tool coverage

Every major distribution, CNI, ingress, security, and operational tool in the Kubernetes ecosystem.

Managed Distributions

EKS (AWS)GKE (Google)AKS (Azure)OKE (Oracle)DOKS (DigitalOcean)Linode LKEScaleway Kapsule

Self-hosted Distributions

k3sRKE2K0sOpenShiftTalos LinuxKubeadmKubesprayMicrok8sKind (dev/test)

GitOps

ArgoCDFlux v2Weave GitOpsArgo RolloutsFlaggerHelmfileTerraform + Helm provider

Packaging

HelmKustomizeHelmfileJsonnetCUEko (Go images)Skaffold

Networking (CNI)

CiliumCalicoFlannelWeave NetAWS VPC CNIGKE Dataplane v2Antrea

Ingress & Gateway

Nginx IngressTraefikHAProxy IngressEnvoyContourIstio GatewayKongGateway API

Service Mesh

IstioLinkerdConsul ConnectCilium Service MeshKumaAWS App Mesh

Security

KyvernoOPA / GatekeeperFalcoTrivy OperatorKubescapekube-benchPolarisNeuVectorCosignKubelinter

Autoscaling

HPAVPAKEDACluster AutoscalerKarpenterCAST AIGoldilocks

Storage

LonghornRook / CephOpenEBSEBS CSIGCS CSIAzure Disk CSINFS ProvisionerMinIO

Backup & DR

VeleroKasten K10StashLonghorn Backupetcd BackupTrilio

Multi-cluster

ArgoCD ApplicationSetsFlux multi-tenancyLiqoAdmiraltySubmarinerCilium Cluster Mesh

Developer Experience

SkaffoldTelepresenceDevSpaceTiltOktetoGardenNix + DevShell

Observability

kube-state-metricsnode-exportercAdvisorPrometheus OperatorGrafanaLokiTempoPixieHubble (Cilium)

What you get

Cluster architecture doc: distribution, node groups, networking, multi-AZ layout
Production cluster provisioned (EKS / GKE / AKS / self-hosted) via Terraform
RBAC: namespaces, roles, service accounts — least privilege, multi-team structure
Network policies: default-deny, pod-to-pod, namespace isolation
GitOps: ArgoCD or Flux with App-of-Apps pattern, env promotion workflow
Helm charts for your applications with values per environment
Ingress controller + TLS automation (cert-manager + Let's Encrypt / ACM)
Autoscaling: HPA + VPA + Cluster Autoscaler or Karpenter for node provisioning
KEDA for event-driven scaling (queue depth, custom metrics)
Pod Security Standards (restricted / baseline) enforced via Kyverno
Falco runtime threat detection with alerting
Velero backup schedule + tested restore procedure
Persistent storage setup per cloud (EBS, GCS, Azure Disk, Longhorn)
Resource quotas, LimitRange, priority classes per namespace
Cluster observability: Prometheus Operator, Grafana dashboards, alerts
Image signing enforcement: Cosign + Kyverno admission webhook
Service mesh (Istio / Linkerd) — optional, scoped to need
Multi-cluster setup with ArgoCD ApplicationSets — optional
Full runbooks + team training session
AspectManaged (EKS / GKE / AKS)Self-hosted (k3s / RKE2)
Control plane opsFully managed by cloud providerYou manage etcd, API server, scheduler
CostControl plane ~$70–150/moFree software, your infra cost only
UpgradesOne-click, minimal downtimeManual, requires careful planning
Multi-cloud / on-premCloud-specificRuns anywhere: bare metal, VMs, edge
Support SLACloud provider SLACommunity + commercial (Rancher)
Best forMost teams — less ops overheadOn-prem, edge, cost-sensitive, air-gapped

Common scenarios

Startup going cloud-native

Moving from bare VMs, Heroku, or ECS. We set up the full stack — cluster, GitOps, monitoring, security — and train your team to operate it independently.

Cluster security audit

Existing cluster with unknown security posture. We run kube-bench, Kubescape, Trivy Operator, and Falco — deliver a prioritized remediation report.

Multi-environment GitOps

Separate dev/staging/prod clusters with consistent ArgoCD ApplicationSets, environment promotion workflow, and drift detection.

Zero-downtime migration

Migrate workloads from old cluster to new with blue/green traffic shifting. No 2am maintenance windows.

Cost rightsizing

Cluster over-provisioned but team is unsure what to cut. We run Goldilocks + VPA analysis and deliver concrete right-sizing recommendations.

Multi-tenancy for SaaS

Isolate customer workloads with namespace-per-tenant, network policies, Kyverno guardrails, and resource quotas. Safe and auditable.

Questions

EKS, GKE, AKS, or self-hosted — which one?

Managed control plane (EKS/GKE/AKS) for most teams — you don't want to manage etcd. EKS if you're AWS-first, GKE if you want the most mature managed K8s, AKS for Azure shops. k3s or RKE2 for on-prem or edge. We help you decide and implement.

Do we actually need Kubernetes?

Not always. If you have fewer than 5 services and a small team, ECS or App Runner might serve you better. We'll be honest about this. Kubernetes adds real value at 10+ services, multiple teams, or complex deployment requirements.

ArgoCD or Flux — what's the difference?

ArgoCD has a better UI and is easier for teams new to GitOps. Flux is more Kubernetes-native, lighter weight, and better for multi-tenancy at scale. Both are excellent. We default to ArgoCD for most teams.

Do we need a service mesh (Istio / Linkerd)?

Probably not at first. Service meshes add real value for mTLS between services, advanced traffic management, and fine-grained observability — but they add operational complexity. We recommend starting without one and adding it when you have a specific need.

What's Karpenter and should we use it?

Karpenter (AWS) is a next-gen node autoscaler that provisions exactly the right instance type for pending pods — much faster and more cost-efficient than Cluster Autoscaler. We recommend it for EKS clusters with variable workload patterns.

How do you handle cluster upgrades?

We set up a documented upgrade runbook: test on staging first, upgrade control plane, then node groups one at a time with cordon/drain. Managed clusters (EKS/GKE/AKS) make this much safer. Velero backup runs before every upgrade.

What security standards do you apply?

CIS Kubernetes Benchmark (via kube-bench), Pod Security Standards (restricted mode), network policies default-deny, Falco runtime detection, image signing with Cosign, and Kyverno admission policies. We deliver a Kubescape score before and after.

How long does a full setup take?

Typically 2–4 weeks for a production-ready cluster with GitOps, security, monitoring, and documentation. Complex setups (multi-cluster, service mesh, multi-tenancy) take 4–8 weeks.

Ready to get started?

Book a free 30-min call. We'll review your current setup (or start from scratch) and give you an honest recommendation.

Book Discovery Call

30 min · Free · No commitment