Long-running workloads on Azure, for less
I architected the Azure services behind an oil & gas product suite around a deterministic orchestration model — and brought the cloud bill down without touching availability.
- Role
- Systems Architect
- Company
- Critical Control Technologies
- Stack
- .NET · Azure
- When
- 2026
The challenge
What needed solving
Production accounting and measurement involve long-running work that must finish correctly even when something fails along the way. The platform also needed one identity strategy across all products, and infrastructure that cost what it should.
What I did
My approach
- Orchestration: Built services on App Services and Durable Functions with a deterministic orchestration model for long-running workloads.
- Messaging and real time: Used Service Bus to decouple work and SignalR to push live updates to users.
- Identity: Evaluated self-hosted and managed identity platforms and integrated Microsoft Entra ID as an external identity provider.
- Cost: Rightsized App Service Plans and consolidated SQL Server VMs.
Deliverables
The work behind the result
Service architecture
App Services, Durable Functions, Service Bus and SignalR working as one platform.
Orchestration model
A deterministic model that makes long-running workloads predictable.
Identity strategy
One approach for the whole product suite, with Entra ID integration.
Cost optimization
Right-sized plans and fewer SQL Server VMs.
Results
What changed
Azure costs
on availability
identity strategy for the product suite