A quick feasibility setup for data transfers between AWS S3 and GCP storage buckets using IPv6.
Some background and general information on IPv6.
See Phase 1
Minimal feasibility / POC setup between AWS S3 -> GCP storage buckets.
This is the most fundamental unknown — does data actually flow over IPv6 between AWS and GCS without touching NAT? Everything else is moot if this doesn't work.
See Phase 2
Measure Transfer Performance and Cost Once the path is confirmed, we could try to create real numbers to justify the approach.
What to do:
- Transfer a meaningful amount of data — to get across any free tier and to emulate real dataset volumes
- Compare against a baseline transfer via IPv4 from a NAT-gated subnet
- Record and Check:
- throughput & transfer time -> performance gains?
- billing -> cost savings?
- critically — check AWS bill for NAT Gateway (should be $0 for IPv6)
- Note: NAT running cost will apply either way, so compare data NAT processing cost only when comparing IPv4 to IPv6
Investigate integration options for the steps-copier. What changes are required on the copier itself? What changes would be required on the hosting infrastructure? Would that fit into current usage scenarios?