Multi-homing

An operator facing multi-homing should ask why the additional service is useful. It might reach customers absent from the first route, perform another task or offer a credible fallback. Improving the first route can be an appropriate response when a missing capability is binding. Restricting participation instead changes the participant’s choices and may create a separate contractual or regulatory question. Neither response should be justified by account totals alone.

CMA Table I.2 reports multi-cloud estimates of 6.8%, 7.2%, 7.4% and 7.5% without spend weighting for 2020–2023; the spend-weighted estimates are 37.6%, 38.5%, 40.0% and 40.8%. These are the report’s estimates under its customer-spend method, not precise population shares. The contrast shows that counting customer records and weighting their spend answer different questions. It does not prove that a customer can move a particular workload.

The same study cautions that name matching, spending thresholds, credits, missing providers and grouped subsidiaries affect the proxy. A customer appearing in several provider records can still face a difficult migration. A useful operating review therefore follows a real workload: what data must be moved, which interfaces must be changed, who can authorize the move and whether the alternative can run the task.

A backup that cannot complete the task is a weak alternative even when an account exists. Consider a business that can log into another service but has not configured the necessary data or permissions. That hypothetical arrangement should not be counted as a demonstrated migration capability.

Different services can be complementary rather than competing. Define the category narrowly enough to identify a real substitution option. Also separate simultaneous participation from a sequence in which the old service is abandoned before the new one is used. Neither broader vendor diversity nor successful switching proves ongoing multi-homing.