On-Premises Integration

Learn how to configure Kubeadapt for on-premises Kubernetes clusters with custom resource pricing for accurate cost tracking.

Kubeadapt supports on-premises deployments by allowing you to configure custom pricing for compute, memory, storage, and network resources. This is ideal for:

• Self-managed Kubernetes clusters (kubeadm, kops, Rancher, etc.)
• Bare metal infrastructure
• Private datacenters
• Hybrid cloud environments

• On-premises Kubernetes cluster (1.24 or higher)
• kubectl access to your cluster
• Internal resource pricing information (CPU, RAM, storage costs)

Kubeadapt uses default pricing based on GCP us-central1 rates if no custom pricing is configured:

json
Copy
1{
2  "provider": "custom",
3  "description": "Default prices based on GCP us-central1",
4  "CPU": "0.031611", // $ per core-hour
5  "spotCPU": "0.006655", // $ per core-hour (spot/preemptible)
6  "RAM": "0.004237", // $ per GB-hour
7  "spotRAM": "0.000892", // $ per GB-hour (spot/preemptible)
8  "GPU": "0.95", // $ per GPU-hour
9  "storage": "0.00005479452", // $ per GB-hour
10  "zoneNetworkEgress": "0.01", // $ per GB
11  "regionNetworkEgress": "0.01",
12  "internetNetworkEgress": "0.12"
13}

To reflect your actual datacenter costs, configure custom pricing in your Helm values file.

Step 1: Calculate Your Resource Costs

Determine your internal costs for:

1. CPU: Cost per core-hour (total server cost / cores / hours per month)
2. RAM: Cost per GB-hour (total memory cost / GB / hours per month)
3. Storage: Cost per GB-hour (storage infrastructure cost)
4. Network: Egress costs (if applicable)

Example calculation:

text
Copy
1Server: $3,000 (24 cores, 128GB RAM)
2Lifespan: 36 months
3Operating hours: 720 hours/month
4
5CPU cost = $3,000 / 36 / 720 / 24 = $0.00579 per core-hour
6RAM cost = $3,000 / 36 / 720 / 128 = $0.00108 per GB-hour

Step 2: Update Helm Values

Create or update your values.yaml file:

yaml
Copy
1# values.yaml
2opencost:
3  opencost:
4    customPricing:
5      enabled: true
6      costModel:
7        description: "Internal datacenter pricing"
8        CPU: "0.00579" # Your calculated CPU cost per core-hour
9        RAM: "0.00108" # Your calculated RAM cost per GB-hour
10        storage: "0.00010" # Your storage cost per GB-hour
11        GPU: "1.50" # GPU cost per hour (if applicable)
12        zoneNetworkEgress: "0.01"
13        regionNetworkEgress: "0.01"
14        internetNetworkEgress: "0.12"

Step 3: Apply Configuration

Install or upgrade Kubeadapt with your custom pricing:

bash
Copy
1helm upgrade kubeadapt kubeadapt/kubeadapt \
2  --namespace kubeadapt \
3  --install \
4  -f values.yaml

For organizations with different server types or pricing tiers, you can set node-specific pricing using Kubernetes labels.

Label Nodes by Pricing Tier

bash
Copy
1# High-performance servers
2kubectl label nodes node1 node2 node3 pricing-tier=premium
3
4# Standard servers
5kubectl label nodes node4 node5 node6 pricing-tier=standard

Configure Tier-Specific Pricing

yaml
Copy
1# values.yaml
2opencost:
3  opencost:
4    customPricing:
5      enabled: true
6      costModel:
7        # Default pricing (standard tier)
8        CPU: "0.00579"
9        RAM: "0.00108"
10
11        # Premium tier pricing (via labels)

Note: Full multi-tier pricing support requires additional configuration. Contact support for advanced pricing models.

Issue: Costs seem too high/low

• Verify your cost calculations include all infrastructure expenses
• Check that cost units are correct (per hour)
• Ensure depreciation period matches your accounting standards

Issue: Custom pricing not applied

• Verify Helm values are correct: kubectl get configmap -n kubeadapt
• Check kubeadapt-opencost logs: kubectl logs -n kubeadapt deployment/kubeadapt-opencost-exporter
• Restart kubeadapt-opencost pods: kubectl rollout restart deployment -n kubeadapt

Issue: Different node types need different pricing

• Use Kubernetes node labels to categorize nodes
• Consider multiple kubeadapt deployments for complex pricing models
• Contact support for enterprise multi-tier pricing solutions

1. Regular Review: Update pricing quarterly to reflect infrastructure changes
2. Include All Costs: Factor in power, cooling, networking, and maintenance
3. Document Assumptions: Keep calculation methodology documented for audits
4. Start Conservative: Begin with estimates, refine over time with actual data
5. Monitor Trends: Use Kubeadapt to identify cost optimization opportunities

• Cost Attribution - Understand how costs are calculated
• Available Savings - Identify optimization opportunities
• Right-sizing - Optimize resource requests