CONCEPT Cited by 1 source
Cross-cloud cost consolidation¶
Cross-cloud cost consolidation is the re-platform strategy of moving components out of a "component A on cloud X, component B on cloud Y" topology and into a single-cloud topology. The savings come from three sources:
- Egress-fee elimination — every byte moved between clouds costs $ per GB (typically $0.08–$0.12/GB for most major providers); same-cloud same-region is free or near-free.
- Operational complexity reduction — two IAM models, two observability stacks, two networking layers, two billing accounts, two sets of availability-zone topologies to reason about collapses to one.
- Engineering-time reduction — on-call shape simplifies, runbooks collapse, the "is it broken in cloud X or cloud Y?" triage step disappears.
Canonical case¶
Segment's 2024-08-01 objects-pipeline post cites consolidation as a co-equal driver alongside the direct cost delta: "We recently revamped our platform by migrating from BigTable to offset the growing cost concerns, consolidate infrastructure to AWS, and simplify the number of components." (Source: sources/2024-08-01-segment-0-6m-year-savings-by-using-s3-for-change-data-capture-for-dynamodb)
Segment's V1 topology ran DynamoDB on AWS (authoritative state) and Bigtable on GCP (CDC changelog) — every changelog read from warehouse- integration infrastructure on AWS paid GCP→AWS egress on the data-transfer leg. V2 consolidated onto AWS by moving the changelog to S3, collapsing the egress line to zero and simplifying the operational footprint. Aggregate savings: ~$0.6M/year.
Structural implication¶
When a cross-cloud topology exists because "we picked the best product per component at the time", the eventual consolidation is often driven not by either product becoming inferior but by the cross-cloud tax growing faster than the per-product advantage. The tax is roughly linear in data-transfer volume, so it compounds at scale — which is why the sign typically flips at petabyte-scale workloads where the cross-cloud egress bill is its own material cost centre.
Seen in¶
- sources/2024-08-01-segment-0-6m-year-savings-by-using-s3-for-change-data-capture-for-dynamodb — canonical wiki instance; Segment consolidates the CDC-changelog component off GCP and onto AWS (S3), citing cost + component-count reduction + simplification all in one paragraph.