Cloud cost estimation
What a system costs is a design decision, not a billing surprise. Skeema maps real cloud SKUs to every node and projects a monthly bill — so you can compare architectures on price the same way you compare them on latency.
What drives a cloud bill
Most cloud spend comes from four buckets:
- Compute
- Running services — instances, containers, or serverless invocations. Usually the largest line item.
- Storage
- Disk and database storage, billed per GB-month, plus IOPS on some tiers.
- Egress (data transfer out)
- Bytes leaving the cloud to the internet. Famously expensive and easy to forget — a key reason to weigh a CDN or an egress-free provider.
- Requests / messages
- Per-operation charges — API Gateway requests, queue messages, function calls.
SKUs and sizing
A SKU is a specific purchasable unit — say an db.r6g.large Postgres instance or an S3 storage class. Skeema maps each node to a realistic SKU from 2026 AWS, GCP, Azure, and self-hostedpricing. Pick a traffic preset (Startup, Growth, Scale, Enterprise) and Skeema auto-sizes each node — bumping instance tiers and replica counts to match the load.
Traffic-variable costs
Some costs only appear under load. Skeema models the big ones:
- •Egress = data-out (GB) × the provider’s per-GB rate.
- •DB storage = stored GB × rate.
- •Queue = RPS × seconds/month ÷ 1M × per-message rate.
Insights
Beyond a number, Skeema surfaces auto-insights — e.g. “compute dominates this bill; Reserved Instances would save ~30%” or “egress is your second-largest cost; a CDN would cut it.” These point you at the line items worth optimizing first.
- ✓Cloud cost = compute + storage + egress + requests. Egress is the sneaky one.
- ✓Skeema maps real SKUs to nodes and auto-sizes them by traffic preset.
- ✓Override SKUs per node; insights point you at the biggest savings.
- ✓Treat the figure as directional — great for comparing designs, not a guaranteed invoice.