π Send over the wire β POST api/transfer with method=wire. The backend computes days = (PBΓ1024) Γ· (effective Gbps Γ· 8 Γ 86,400 Γ· 1024) and compares against the deadline. The dashed wire animates; the progress bar fills proportionally to the calculated days.
π Ship by truck β method=truck with your device choice. The backend validates you ordered enough devices (Snowmobile = 100 PB, Data Box Heavy = 1 PB) and returns the five-stage logistics timeline. The truck animates: arrival at the data centre β 3-day ingestion pause β transport β cloud ingestion.
Sliders change the inputs; invalid combinations (e.g., 2 Data Box devices for 5 PB) return an HTTP 422 which is shown as an error.
5 PB Γ· 75 TB/day β 68 days β misses a 60-day target even with perfect uptime.
Truck: order β 3-day local ingest β ship β cloud ingest β 18 days.
Why
Bandwidth is shared with production replication and users.
Shipping 100 PB in a truck is still faster than most networks (βnever underestimate the bandwidth of a truckβ).
When
Bulk historic data, backups, media archives β truck.
Active databases with constant change β wire, with continuous replication.
Pitfalls
Ignoring the lead time to provision Direct Connect/ExpressRoute (weeks).
Forgetting that data keeps changing while the truck drives β plan the delta.
Not validating checksums before wiping the source.
Key ideas
Hybrid is the real answer
Truck for the 5 PB bulk, wire for the final delta. The cutover page assumes exactly this: a last block sync at Saturday 2 AM.
Constraints
Snowmobile requires site survey, power and security escort. Data Box Heavy requires 2 Γ 40 GbE and a pallet jack. Both need a loading dock booking β logistics, not technology.
Gbps β TB/daySnowmobileData Box HeavyChain of custodyDelta sync