Cloud landing zones
Shared cloud services need an explainable cost model
Central networking and observability can improve consistency, but their costs still belong somewhere. Define the allocation before teams depend on them.
In this article
Identify the shared bill
List platform services used by several workloads, such as connectivity, security monitoring and deployment infrastructure. Separate fixed baseline cost from usage that grows with traffic or data volume.
A workload's own resource bill may exclude a substantial share of its operating cost. Showing only that direct bill can make architecture choices look cheaper than they are.
Use actual provider billing data for estimates and identify currencies and billing periods clearly. Avoid comparing totals that cover different time windows.
Choose a rule teams can understand
Some shared costs may remain a central platform budget. Others may be allocated using a defensible usage measure or a simple agreed split.
Explain the limitations. Log volume may be a reasonable driver for one service, while network transfer or reserved capacity may drive another. A precise-looking allocation is not necessarily a fair representation of cause.
Keep the rule stable enough for planning and review it when the platform changes. Teams should not discover an unexplained shared charge after making a deployment decision.
Include the cost of complexity
Creating many isolated environments can improve boundaries but also duplicate baseline resources and support work. Centralising everything can reduce duplication while increasing shared dependencies and coordination.
Compare the operational consequences alongside the bill. A shared service that requires frequent manual changes for every workload may consume more engineering time than expected.
Use a representative workload to measure both resource consumption and support effort before expanding the pattern broadly.
Set ownership and review points
Give platform owners visibility into total shared spend and workload owners visibility into the portion they can influence. Alerts should reach someone able to act on the relevant driver.
Track unused environments and resources with a removal process that respects data and recovery needs. A missing owner tag should trigger investigation rather than automatic destructive cleanup.
The cost model is useful when teams can explain why a service costs what it does and how a design change would affect that cost. Allocation should support decisions, not merely redistribute a bill until every row has a number.
Primary sources
Microsoft: Azure landing zonesReferences checked 11 September 2026.