Professional scenario conversions
Cloud billing units explained
Cloud bills combine compute hours, storage gigabyte-months, request counts, and egress. Translating line items into product metrics needs explicit unit dictionaries shared by finance and engineering.
Credits, taxes, and enterprise discounts apply in invoice order—unit economics spreadsheets should mirror that sequence or they will show phantom savings.
Key takeaways
- Separate list prices from blended effective $/vCPU·h after commitments.
- GB·month is prorated—confirm vendor rules for daily vs monthly averages.
- Egress often bills as multiple products—map each line to architecture.
- Tags make attribution possible; untagged spend is only a blended average.
How to convert
100 GB = 102,400 MB
vCPU-hour and memory-hour
Billed time granularity may be per second or per hour with minimums—read the pricing page’s rounding rules.
GB-month for disks and object storage
Average daily usage over a month can differ from peak-day usage; cost models may prorate differently.
Egress vs ingress
Data leaving a region often costs more than data entering—unit price is not symmetric.
From bill to unit economics
Divide monthly spend by active users, requests, or GB stored—pick the denominator that matches your pricing strategy.
Tagging and accountability
Consistent resource tags make attribution possible; without tags, engineers see only a blended average.
Reserved capacity vs on-demand
Commit discounts change the effective $/vCPU-hour; compare blended rates, not list prices, when pitching autoscaling.
Cross-region replication
Replication bytes may bill as multiple products—snapshot shipping plus continuous sync—map each line item to architecture diagrams.
Credits and enterprise agreements
Private pricing and credits apply in invoice order; unit economics spreadsheets should mirror that sequence to avoid phantom savings.