Dated starting points only — Unknown until verified means no invented $/GB. PipeToll does not scrape private consoles or guarantee figures.
This spoke is a path matrix for object-storage egress on AWS S3 versus Google Cloud Storage. Titles lead with Cloud egress so readers are not hunting American pipe-toll brands. PipeToll refuses invented list prices — cells stay Unknown until verified.
Path templates to model
Client ← Internet ← S3/GCS — classic public GET; free grants and tiered Internet download meters apply after cliffs.
VPC/VPC-SC compute ← private path — gateway/interface endpoints or Private Google Access / PSC shift meters; watch private-path fees.
CRR / multi-region replication — bytes that never touch an end user still bill; keep separate from user downloads.
Client ← CDN ← object origin — miss-driven origin fill may sit on storage or CDN bills depending on pairing.
Analytics export S3 ↔ GCS — the source usually pays egress; dual-running months understate cutovers if ignored.
Common traps
Pasting one blog “egress $/GB” across clouds ignores free-tier cliffs, Class A/B or GCS ops classes, and CDN origin bills. Treating multi-region as “free HA” hides replication meters. Assuming private endpoints zero every line item is unsafe. No scorecrowns from Incomplete cells — badge Unknown. No guaranteed savings from switching logos.
FinOps checklist
Inventory top prefixes by GET volume and bytes on each cloud.
Separate Internet, private, replication, and CDN-origin lines.
Cite dated provider pages per methodology; share vocabulary via the glossary.
FinOps-liteUnknown until verified
Worked example (educational): 6 TB public GET + 800 GB replication
Path A: clients download 6,000 GB from object storage over Internet. Path B: 800 GB of cross-region / multi-region replication in the same month. Run path templates separately in Mode A — do not blend replication with user egress. Paste dated $/GB or leave cells Unknown. Not a quote or migration savings guarantee.
