Validated on 2 Mar 2023 • Last edited on 1 Oct 2024
DigitalOcean Kubernetes (DOKS) is a managed Kubernetes service. Deploy Kubernetes clusters with a fully managed control plane, high availability, autoscaling, and native integration with DigitalOcean Load Balancers and volumes. You can add node pools using shared and dedicated CPUs, and NVIDIA H100 GPUs in a single GPU or 8 GPU configuration. DOKS clusters are compatible with standard Kubernetes toolchains and the DigitalOcean API and CLI.
Node Pools
The total cost of a DOKS cluster depends on the configuration and usage of its node pools throughout the month. Node pools can have a variable number of nodes over the course of a month due to auto-scaling or manual changes.
Worker nodes are built on Droplets and are charged per second at the same price as Droplets. Billing starts when the node in the node pool is ready, even if it is unhealthy. We do not charge for created nodes that do not join the cluster.
At minimum, we charge node pools for 60 seconds of usage or $0.01 per node in the pool, whichever is greater. At most, we charge for 28 days of usage per month. If there are more than 28 days in the month, we charge for the 28 days (672 hours) with the highest usage, and don’t charge for any additional usage.
Click for more details and examples of node pool pricing.
Worker nodes are charged per second.
For example, consider a cluster with:
Three worker nodes that cost $100/month ($0.00004134/second)
No autoscaling
27 days, 1 hour, and 20 seconds (2,336,420 seconds) of usage
The total cost would be 3 * $0.00004134/second * 2,336,420 seconds = $289.73.
At minimum, we charge node pools for 60 seconds of usage or $0.01 per node in the pool, whichever is greater. For node pools with varying number of nodes with different time spans, we aggregate the usage first.
For example, if a worker pool has 3 nodes with 20 seconds of usage each earlier in the month, then 15 seconds of usage later in the month, the aggregate usage is 35 seconds. For billing, we round this usage up to 60 seconds. If that results in a bill of less than $0.03 ($0.01 per node), we round the bill up to $0.03.
At most, we charge node pools for 28 days of usage per month. If there are more than 28 days in the month, we charge for the 28 days (672 hours) with the highest usage, and don’t charge for any additional usage.
For example, consider a cluster that:
Runs continuously through a month with 31 days (744 hours)
Has a node pool that varies between 10 and 25 nodes
Uses worker nodes that cost $100/month ($0.00004134/second)
A node pool’s fixed size over a time is called a span. To calculate the total cost, we aggregate node pool spans with the same number of nodes over the month, and sort in descending order:
Number of Nodes
Node Span (Hours)
Span Cost
25
100
$0.00004134 * 25 * 100 * 3600 seconds = $372.06
20
100
$0.00004134 * 20 * 100 * 3600 seconds = $297.65
15
100
$0.00004134 * 15 * 100 * 3600 seconds = $223.24
15
100
$0.00004134 * 15 * 100 * 3600 seconds = $223.24
15
100
$0.00004134 * 15 * 100 * 3600 seconds = $223.24
10
100
$0.00004134 * 10 * 100 * 3600 seconds = $148.82
10
100
$0.00004134 * 10 * 72 * 3600 seconds = $107.15 Reached billing cap of 672 hours. No billing for 28 hours in span.
10
44
No billing for 44 hours in span.
The 672 hours billing cap is reached in the second span with 10 nodes. While the usage is for a total of 744 hours (31 days), we bill you only for 672 hours (28 days). Thus, the total node pool cost is $372.06 + $297.65 + $223.20 + $223.24 + $223.24 + $148.82 + $107.15 = $1595.28.
GPU Worker Nodes
GPU worker nodes are priced per second at the same price as the GPU Droplets. For reservation and contract pricing, contact your sales representative or Customer Success Manager, or send a request using the H100 GPU Worker Nodes form.
Control Plane and High Availability
The control plane is fully managed by DigitalOcean and included at no cost.
DOKS clusters accrue free bandwidth based on the worker pool’s largest sizes within 28 days of usage.
For example, if your worker pool has Droplets with a monthly bandwidth quota of 5TB, then you accrue free bandwidth at the rate of 5TiB/(24*28) * worker_pool_size = 7.44 GiB/hr * worker_pool_size. For autoscaling clusters, the size of the worker pools vary.