Containers
Managed Kubernetes on dedicated worker nodes
Production-ready clusters built on Cluster API — self-healing nodes, autoscaling worker pools and a highly available control plane, deployed in minutes and managed for you.
Athens · Barcelona · BostonUnlimited bandwidthCluster ready in minutes
portal.antyxsoft.io
Container orchestration, fully managed
Upstream Kubernetes with the lifecycle work handled for you — provisioning, healing, scaling and upgrades driven declaratively through Cluster API.
Container orchestration
Scheduling, placement and load balancing of your containers across the cluster, handled automatically.
Automated scaling
Cluster Autoscaler adjusts worker counts to pending pods, on resource utilisation or your own custom metrics.
Declarative configuration
Describe the desired state in YAML; the cluster reconciles the actual state to match it continuously.
Storage orchestration
Persistent and ephemeral volumes provisioned on demand, so stateful workloads run as first-class citizens.
Rolling updates and rollbacks
Ship new versions with no disruption, and roll back to the previous revision the moment something looks wrong.
Service discovery and load balancing
Built-in DNS and service load balancing expose your workloads without manual network plumbing.
Cluster provisioning
From cluster creation to first deployment in minutes
Choose a region, a Kubernetes version and your node pools, and a fully conformant cluster is provisioned with kubeconfig ready to download — no control-plane assembly, no etcd to babysit.
Upstream Kubernetes releases with a managed upgrade path
kubectl, Helm and Terraform supported out of the box
Node pools resized or replaced without cluster downtime
Create cluster
ParameterStatus
v1.31.2
Kubernetes version
Athens · eu-ath-1
Region
3 × CS4 — 4 vCPU / 8 GB
Worker pool
kubeconfig.yaml
Cluster access
Provisioned in 4m 12s — no control-plane assembly, no etcd to babysit.
Operations
Cluster operations without the operational overhead
Inspect nodes, workloads, volumes and ingress from the same panel as the rest of your infrastructure, with node health and consumption visible in real time — and one invoice for the whole estate.
Node health, capacity and event history per pool
Namespace and project isolation for teams and environments
Flexible billing — prepaid credit or monthly invoicing
Node pools
PoolHealth
pool-general
6 nodes · 24 vCPU · 48 GB
pool-memory
3 nodes · 12 vCPU · 96 GB
pool-batch
2 → 4 nodes · scaling out
pool-edge
1 node · drained for upgrade
Node health, capacity and event history per pool — one invoice for the whole estate.
Resilience
Self-healing clusters, down to the infrastructure
Cluster API manages the infrastructure and Kubernetes together, so a failed node is not just rescheduled — it is replaced. Machine HealthCheck certifies control-plane and worker health, provisions a replacement node and lets Kubernetes reschedule the affected pods onto it.
Automatic node replacement on failure or resource exhaustion
Health monitoring across control-plane and worker nodes
Highly available control plane on custom plans
Machine HealthCheck
EventElapsed
pool-general-3 stopped reporting
kubelet
Marked unhealthy, drain started
Machine HealthCheck
Replacement node provisioned
Cluster API
Pods rescheduled, pool at capacity
kube-scheduler
A failed node is replaced, not just rescheduled.
Standardisation
One lifecycle model across every cluster
Cluster API abstracts the deployment mechanics of each provider, so teams running heterogeneous infrastructure standardise cluster lifecycle management regardless of vendor, data centre or edge site — and reuse the same components across clusters instead of rebuilding them.
Consistent provisioning and upgrade workflow per cluster
Component reuse across environments, reducing rework
Declarative expansion of control-plane and worker pools
Cluster inventory
ClusterState
prod-athens
v1.31.2 · 9 nodes · Athens
prod-barcelona
v1.31.2 · 6 nodes · Barcelona
staging-athens
v1.31.2 · 3 nodes · Athens
edge-boston
v1.30.6 · 2 nodes · Boston
One declarative lifecycle model, regardless of vendor, data centre or edge site.
Kubernetes cluster specifications
Processor
Dedicated vCPU workers
Bandwidth
Unlimited
Node memory
2 GB – 96 GB
Regions
Athens, Barcelona, Boston
Node storage
40 GB – 400 GB NVMe
Billing
Monthly or annual
Outbound network
Up to 10 Gbps
Add-ons
Registry, VPC, LB, CSI
Kubernetes node plans
Pre-configured plans you can deploy from the portal in a few clicks — or build a custom plan with your own resources, node count and control-plane high availability.
Plan
vCPUs
RAM
Node storage
Bandwidth
Monthly
How managed Kubernetes works at Antyxsoft
Four steps in the portal, then your own toolchain takes over — kubectl, Helm and Terraform all talk to the cluster the moment it is up.
-
1
Choose a plan and a region
Pick a pre-configured plan or build a custom one — resources, node count and control-plane high availability are yours to set. -
2
Define your node pools
Separate pools per workload class, each with its own size and autoscaling bounds. -
3
Download the kubeconfig
One file, immediate cluster access. No bastion to configure, no VPN to join. -
4
Deploy and let it scale
Apply your manifests; the autoscaler adds and removes workers as pending pods demand.
zsh — antyx-prod
$export KUBECONFIG=~/antyx-prod.yaml
$kubectl get nodes
NAME STATUS ROLES VERSION
antyx-cp-1 Ready control v1.33.2
antyx-pool-a1 Ready worker v1.33.2
antyx-pool-a2 Ready worker v1.33.2
$helm install api ./charts/api
STATUS: deployed
What teams run on Antyxsoft Kubernetes
The same cluster model serves a two-service API and a multi-tenant platform — you change the node pools, not the architecture.
Service-oriented applications
Dozens of independently deployed services with internal discovery, per-service scaling and no manual load-balancer wiring.
Multi-tenant products
Namespace and quota isolation per tenant or environment, with rolling releases that never take the platform offline.
Build and test pipelines
Ephemeral runners that scale to queue depth and disappear when the queue clears, so you stop paying for idle build capacity.
Scheduled and queued processing
CronJobs and job queues on pools sized for throughput, with persistent volumes for intermediate state.
Inference and training workloads
Serve models on dedicated-vCPU pools, or attach GPU nodes for training and batch inference in the same cluster.
Lift from EKS, AKS or GKE
Standard upstream Kubernetes with no proprietary control-plane extensions — your manifests and Helm charts move as they are.
Cost comparison
Kubernetes cost comparison: the same cluster, without the hyperscale premium
Flat node pricing and unlimited bandwidth mean a cluster costs what its nodes cost — no per-cluster management fee, no egress line, no reserved-capacity commitment.
Calculate your saving3 nodes · 4 vCPU / 8 GB each · control plane · 4 TB egress
Antyxsoft
159,00 €
GCP
≈ 322 €
Azure
≈ 371 €
AWS
≈ 428 €
Illustrative list prices for a comparable configuration including outbound transfer.
Ready to deploy? The docs have the exact steps.
Creating a cluster, scaling node pools, configuring ingress and attaching persistent volumes — all in the knowledge base.
Frequently asked questions about Antyxsoft Kubernetes
What is Antyxsoft Kubernetes?
A managed platform for running containerised applications: it automates deployment, scales your workloads and keeps them healthy across environments, with the cluster lifecycle handled through Cluster API.
Is the control plane managed for me?
Yes. Antyxsoft runs the control plane, including upgrades and high availability, so your team only manages workloads and node pools.
How does cluster autoscaling work?
Node pools scale between the minimum and maximum you set, adding nodes when workloads are unschedulable and removing them when capacity is idle.
Can clusters use private networking?
Yes. Clusters can be attached to a Virtual Private Cloud so nodes and services stay on isolated networks, with firewalls controlling ingress.
Guides for running Kubernetes in production
Visit the blog →
Networking · 7 min read
Ingress, LoadBalancer or NodePort
Which of the three to use for each service — and why one load balancer per service gets expensive.
Storage · 8 min read
Persistent volumes: what to use for what
Block, file and object storage in a cluster, plus the reclaim policy that quietly deletes production data.
Operations · 8 min read
Upgrading a production cluster without downtime
Pre-flight checks, PodDisruptionBudgets and the upgrade order — with no rollback for a control plane.
Deploy your first cluster today
Pick a region, size your node pools, and hand kubectl a running cluster.