+1 (201) 771-1665 Sales Support 24/7 Engagements active in 5 regions
Field Notes

Multi-region Cost Model -- Decisions & Trade-offs | Cobham Consulting Group

Multi-region cloud costs are consistently underestimated. The bandwidth charges for data replication, the standby capacity required for failover, and the latency tax of distributed consensus add up to a bill that surprises most organizations the first time they see it.

The components most often missed

Cross-region egress is charged by most providers at 2 to 9 cents per GB. A database that replicates 500GB per day across regions generates 15,000 dollars or more per month in egress before any application traffic. This number almost never appears in initial architecture estimates because it requires knowing replication volume, which requires running the system first.

Standby capacity in a second region that runs at 50 percent scale for failover readiness adds 50 percent to your compute bill for infrastructure that serves no traffic under normal conditions. Active-active designs avoid idle capacity but require your application to handle split-brain scenarios, which adds engineering complexity that must be maintained.

The break-even analysis

Active-active multi-region is justified when the cost of downtime -- measured in revenue lost per hour times MTTR -- exceeds the cost of the additional infrastructure over the reliability improvement window. For most workloads, that break-even point is higher than assumed. An active-passive design with a 15-minute RTO is often cheaper and more reliable than an active-active design maintained by a team without deep distributed systems experience.

Run the math before the architecture decision, not after. The inputs are: cost of downtime per hour, expected MTTR with active-passive, expected MTTR with active-active, additional monthly infrastructure cost of active-active, and your required reliability SLA.

Common questions

Frequently asked.

How much does cross-region egress cost on AWS?
AWS charges 0.01 to 0.09 per GB for cross-region data transfer depending on the regions. A database replicating 500GB per day generates 1,500 to 13,500 dollars per month in egress charges before any application traffic. Model this before designing multi-region replication.
What is a reasonable RTO for an active-passive failover?
With pre-warmed instances and automated failover, 5 to 15 minutes is achievable for most database and application workloads. Cold failover without automation can take 30 to 60 minutes. Define your acceptable RTO, then design to meet it -- not the other way around.
Is multi-region worth it for a startup?
Rarely at early stage. The engineering overhead of multi-region consistency exceeds the reliability benefit until traffic is high enough that a single-region outage would be materially damaging. A robust single-region setup with tested backups and a clear runbook is the right default.
All field notes
Questions on this?

Talk to a principal.

If this essay described a problem you are working on, schedule a 30-minute call. We map your specific situation and tell you what we would do first.

Schedule a call
Direct to a principal · Response under four hours