DigitalOcean Kubernetes

by DigitalOcean • • Container Orchestration Tools

No reviews yet
Free plan available View plans

DigitalOcean Kubernetes (DOKS) is a managed Kubernetes service for developers, start-ups and AI teams, running containers on DigitalOcean Droplet worker nodes.

DigitalOcean Kubernetes takes the work of operating a cluster's control plane off a team: DigitalOcean runs and patches it, and the team looks after its own containers. Clusters are managed through the DigitalOcean API, command-line tool or web interface. The service, known as DOKS, belongs to DigitalOcean, the NYSE-listed cloud company founded in 2012. DOKS conforms to the Cloud Native Computing Foundation's Kubernetes Software Conformance Certification program. Every cluster ships with Cilium for overlay networking and network policies, with Cilium Hubble adding security and observability views. CoreDNS gives services stable names, and a Service of type LoadBalancer creates a DigitalOcean Load Balancer in front of the worker nodes. The NGINX ingress controller or the managed Gateway API handles HTTP routing, while persistent volume claims on the do-block-storage class provision Volumes Block Storage. The Cluster Autoscaler adds and removes nodes as pods wait for capacity, and Horizontal Pod Autoscaling adds replicas on sustained CPU spikes. Automatic upgrades apply patch releases inside a chosen maintenance window, and Kubernetes 1-Click Apps install Helm charts from the Marketplace. DOKS suits developers and start-ups that want Kubernetes without running its control plane, and small and medium-sized businesses that already build on DigitalOcean. Worker nodes are Droplets, so a team picks Basic, CPU-optimized, general purpose, memory-optimized or storage-optimized machines for each node pool. The v5 nodes target AI and ML tools, bursty data pipelines and high-throughput services. Teams training or serving models can add NVIDIA H100 GPU nodes. Production clusters can switch on a high availability control plane that runs several replicas of its components. A typical cluster serves web applications and APIs behind a Load Balancer, pulling its images from DigitalOcean Container Registry and keeping state on Block Storage volumes. Shared files can live on NFS storage that the cluster provisions on demand. Delivery runs from GitHub Actions, which pushes each new build to the cluster, and cluster logs are forwarded to OpenSearch for search. For AI work, teams spin up GPU-powered environments in DOKS and scale them along with the rest of their workloads.

Pricing

View plans
  • Control plane

    Free

    • Per cluster, no charge; worker nodes and other resources billed separately
    • Fully managed by DigitalOcean
    • Single replica of control plane components, so an unexpected failure can cause downtime without high availability
    • Cluster autoscaler and version updates at no extra cost
    • Cilium Hubble security and observability at no extra cost
    • Container Registry included up to 500 MiB
    • Block Storage volumes billed separately from $10 a month
    • Load Balancers billed separately from $12 a month per node
  • Basic worker node

    $12 per month

    Per node per month, billed per second, capped at 672 hours a month

    • Shared CPU for variable workloads
    • Inbound data transfer included
    • Outbound transfer from 2,000 GiB a month per node, pooled across nodes, then $0.01 per GiB
    • Minimum charge of 60 seconds or $0.01 per node
  • CPU-optimized worker node

    $42 per month

    Per node per month, billed per second, capped at 672 hours a month

    • Dedicated CPU with 2 GiB RAM per CPU
    • Everything in the Basic node
  • General purpose worker node

    $63 per month

    Per node per month, billed per second, capped at 672 hours a month

    • Dedicated CPU with 4 GiB RAM per CPU
    • Everything in the Basic node
  • Memory-optimized worker node

    $84 per month

    Per node per month, billed per second, capped at 672 hours a month

    • Dedicated CPU with 8 GiB RAM per CPU
    • Everything in the Basic node
  • Storage-optimized worker node

    $163 per month

    Per node per month, billed per second, capped at 672 hours a month

    • Dedicated CPU with guaranteed NVMe, 225 GiB storage per CPU
    • Up to 6.87 TiB of local storage
    • Everything in the Basic node
  • v5 worker node

    Custom pricing

    • From $0.031 per node per hour with 1 vCPU and 2 GB RAM, billed per second with no monthly cap
    • Shared or dedicated CPU on 5th Gen AMD EPYC processors
    • Monthly total varies with the hours in the month
  • NVIDIA H100 GPU worker node

    Custom pricing

    • From $4.41 per node per hour on demand
    • 80 GB GPU RAM, 240 GiB Droplet RAM and 20 Droplet vCPUs
    • 5 TiB NVMe scratch disk
    • Reserved H100 capacity and contract pricing through sales
  • High availability control plane

    $40 per month

    Per cluster per month, prorated hourly, added to a cluster

    • Multiple replicas of control plane components

Features

  • Included: Declarative scheduling with resource requests
  • Not included: Self-healing restart and rescheduling
  • Not included: Rolling updates with rollback
  • Included: Workload autoscaling on metrics
  • Included: Cluster node autoscaling
  • Included: Built-in service discovery and load balancing
  • Included: Ingress and external load balancer integration
  • Included: Dynamic persistent volume provisioning
  • Not included: Config and secret injection into workloads
  • Not included: Namespace isolation with RBAC and quotas
  • Included: Network policies between workloads
  • Included: Managed control plane offering
  • Not included: Multi-cluster management console
  • Included: Automated cluster version upgrades
  • Included: Helm chart application catalog
  • Not included: GitOps reconciliation from a repository
  • Included: CNCF-certified Kubernetes conformance

Additional Features

  • Fully managed Kubernetes control plane
  • High availability control plane option
  • Cluster Autoscaler with scale to zero
  • Horizontal Pod Autoscaling on CPU spikes
  • Cilium overlay networking
  • Cilium Hubble security and observability
  • Managed CoreDNS
  • DigitalOcean Load Balancers for LoadBalancer Services
  • NGINX ingress controller
  • Managed Gateway API
  • Block Storage volumes through the do-block-storage class
  • NFS storage provisioned on demand
  • Automatic patch upgrades in a maintenance window
  • Surge upgrades with extra nodes
  • Kubernetes 1-Click Apps built on Helm charts
  • Worker node pools on Droplet plans
  • NVIDIA H100 GPU worker nodes
  • DigitalOcean Container Registry integration
  • OpenSearch log forwarding
  • GitHub Actions push-to-deploy
  • Management through the DigitalOcean API, CLI and control panel

Best for

  • Developers and start-ups that want Kubernetes without running a control plane
  • Small and medium-sized businesses already building on DigitalOcean
  • Teams spinning up GPU-powered environments for AI and ML work
  • Teams running bursty data pipelines and high-throughput services
  • Production teams that need a replicated control plane
  • Digital native companies scaling container workloads on one cloud account

Use cases

  • Running containerized web applications and APIs
  • Spinning up GPU-powered environments for inference or training
  • Adding pod replicas when CPU use climbs
  • Adding and removing nodes as pods wait for capacity
  • Exposing services through load balancers and ingress rules
  • Attaching block storage volumes to stateful workloads
  • Installing apps from the Kubernetes 1-Click Apps catalog
  • Restricting traffic between pods with network policies
  • Keeping clusters on current patch versions inside a maintenance window
  • Forwarding cluster logs to OpenSearch
  • Deploying new builds from GitHub Actions
  • Automating infrastructure and software delivery

Screenshots & Videos

Explore DigitalOcean Kubernetes in action

Digitalocean Kubernetes Product Page Hero
View full size
Digitalocean Kubernetes Product Page Hero

User Reviews

Write a Review

No reviews yet

Be the first to share your experience with this software.

Compare with Similar Software

Click to compare side by side with DigitalOcean Kubernetes

Want to claim this brand?

To claim this brand, please contact support.