FinOps¶
Three views on money, all deterministic and saved so you (or an agent) can reason about them:
| View | Command | Source | Needs |
|---|---|---|---|
| Estimate | cs finops estimate | cloudseed's own inventory and config at on-demand list prices | nothing: offline, works right after a --dry-run |
| Actual bill | cs finops cloud | AWS Cost Explorer, Azure Cost Management | the cloud CLI and billing read access; GCP needs a BigQuery billing export (cloudseed tells you how) |
| Kubernetes allocation | cs finops k8s | OpenCost | cs platform install finops |
| Everything | cs finops report --save | all of the above | saved to <workdir>/finops/latest.json |
Estimate before you build¶
╭─ Estimate · aws-demo (month, 730h at on-demand list prices) ──────────────╮
│ bastion t3.micro x1 $7.59 │
│ NAT gateway x1 $32.85 │
│ public IPv4 x2 $7.30 │
│ KMS key x1 $1.00 │
│ CloudTrail + logs (low vol.) $3.00 │
│ GuardDuty (low volume) $5.00 │
│ block storage ~10 GB $0.80 │
│ │
│ total / month $57.54 │
│ │
│ VPC flow logs (30-day retention) are billed per GB ingested and stored │
│ in CloudWatch: not included. │
│ Prices are us-east-1 on-demand list prices. Data transfer and NAT │
│ processing are not included. │
╰────────────────────────────────────────────────────────────────────────────╯
Default landing zones, estimated the same way (monthly, on-demand list prices, before data transfer):
| Target | Estimate | Biggest item |
|---|---|---|
| AWS (us-east-1) | about $58 | the NAT gateway ($32.85) |
| Azure (eastus) | about $55 | the NAT gateway ($32.85) |
| Google Cloud (us-central1) | about $11 | the e2-micro bastion ($6.13) |
| VMware | $0 | runs on your machine |
A Kubernetes cluster, a VPN host or FIPS images add to these; run the estimate after changing the configuration. Usage-billed services (GuardDuty, Security Hub, AWS Config, Log Analytics) are priced at a small environment's volume.
The actual bill¶
The bill by service from the provider, for the environment's account or subscription. Compare it with the estimate: a large gap usually means resources outside cloudseed (check cs inventory), data transfer or logging volume.
Kubernetes cost allocation¶
cs platform install finops # OpenCost + Prometheus (+ metrics-server, Goldilocks)
cs finops k8s --by namespace --window 7d
cs finops k8s --by controller --window 24h
Allocation by namespace, controller, pod or label, with an efficiency figure so idle requests stand out. Efficiency below about 30% means over-requested CPU or memory.
Save and ask for savings¶
The cloudseed-finops skill knows the report format and the levers cloudseed offers:
| Lever | How |
|---|---|
| Right-size pods | Goldilocks recommendations (cs platform install goldilocks), then VPA |
| Scale to zero | KEDA for event-driven workloads; kube-green sleep schedules for office-hours environments |
| Fewer or smaller nodes | cs node remove, cs node scale, --var kubernetes_node_size=... |
| One NAT gateway | --var single_nat_gateway=true (the AWS default; per-AZ NAT is more resilient and costs more) |
| One load balancer | the shared Gateway instead of a load balancer per app |
| Drop what is idle | --var enable_vpn=false, --var enable_kubernetes=false, or cs destroy for a whole environment |
Spot capacity
cloudseed creates on-demand nodes only. Spot or preemptible capacity is a change you make outside cloudseed, and the agent labels it that way.
In the web console¶
Each environment card has a Cost button that runs the estimate, and the Agents & MCP view can run the savings task. See Web console.
Related¶
- Scenario 15 - web console and FinOps
cs help finops,cs explain finops