We recently migrated to a multi-cloud strategy using AWS and Azure, but our monthly billing is becoming unpredictable. I am looking for community advice on how to start with FinOps. What are the best practices for tagging resources and setting up automated cost guardrails to prevent budget overruns without slowing down our dev teams?
3 answers
Starting with FinOps requires a cultural shift where finance and engineering speak the same language. You should begin by enforcing a strict tagging policy; if a resource isn't tagged with a cost center or project ID, it shouldn't be deployed. I recommend using tools like AWS Cost Explorer or Azure Cost Management to gain granular visibility. Most organizations find success by moving from monthly reporting to real-time "cost-as-code" guardrails. This ensures that your development teams remain agile while staying within the predefined budgetary boundaries of your organization's cloud strategy.
Are you finding that the complexity of your architecture is the main driver of these costs, or is it more about orphaned resources and "zombie" instances that aren't being shut down after use?
Focus on identifying your "unit metrics," like cost per request or cost per GB, to truly understand the ROI of your cloud spend rather than just looking at the total bill.
Exactly, Sarah! Identifying the unit cost is the only way to see if your scaling is actually efficient or just wasteful growth.
It's actually a mix of both. We have several test environments that stay running 24/7, and our data egress fees between AWS and Azure are higher than we anticipated. We really need a way to automate the shutdown of non-production instances and optimize our cross-cloud data transfer routes to keep the overhead manageable.