Lely | Tom Kennes

Lely

· 2 min read · 334 words · Skyworkz Lely Azure Cloud Costmanagement KEDA

Through Skyworkz as an external consultant.

 

Context

Lely, a global leader in agricultural robotics, was scaling its Azure footprint across product development and innovations. The central Azure team needed support managing cloud resources, cost optimization, and data infrastructure. Simultaneously, management had expressed interest in establishing sustainable IT practices across the organization.

 

Intervention

Operated in a dual role: supporting the central Azure team on cloud engineering (networking, resource sharing, cost management, KEDA on Kubernetes) while independently leading sustainable IT initiatives. Organized a digital cleanup day to identify sustainability advocates, tested carbon footprint measurement tools, and established partnerships with the sustainability office.

 

Business Impact

  • Deployed KEDA on Kubernetes, providing function-as-a-service capabilities that reduced infrastructure overhead for development teams
  • Implemented Azure cost ingestion and reporting, enabling data-driven cost optimization decisions
  • Led migrations from legacy tenant to new Azure environment, improving operational baseline
  • Established the Microsoft Azure Emissions Impact dashboard, creating organizational visibility into IT carbon emissions for the first time
  • Published detailed analysis of IT emissions measurement, contributing to the broader sustainability community
 

Technology Stack

For my work with the central Azure team, the most important central components were obviously Azure, terraform, Kubernetes, ArgoCD and Gitlab. Within Azure I was working with AKS, EventHubs, various database solutions, various storage solutions, Eventgrid, Data Factory, Azure Functions, Azure App Services and various aspects of networking. Next to that, I’ve setup KEDA on top of AKS and provided the developers with containers to run their python code. And within the migrations I was involved with a lot other Azure components around monitoring.

For the costmanagement analyses, I used DataBricks, PySpark, and the new Lakeview dashboards.

And for the sustainable IT efforts I spent a lot of effort in deciphering the results from the Cloud Carbon Footprint, and identifying its shortcomings. For some of those I did open Pull Requests, but other issues such as missing new Azure services, SKUs and instance types were sometimes more concerning. All in all, it remains tough to assess the footprint behind your cloud resources.