I'm currently reviewing an Azure bill for a client looking for cost savings, and I've noticed a lot of potential opportunities. While rightsizing, utilizing reservations, and using Dev/Test subscriptions for non-production resources seem like solid strategies, there's a significant number of 'zombie' resources here. These are resources that were created for a legitimate purpose but are no longer needed. Individually, they don't cost much, but together they're adding up.
I've suggested standard FinOps practices like assigning owners to Subscriptions and Resource Groups. However, I'm struggling with how to effectively identify these zombie resources. What's a good approach for determining which resources are no longer necessary? Should there be regular meetings to review resources? Would tagging help for periodic check-ins? How can we leverage resource utilization metrics?
Most of these resources are active, so while identifying orphaned ones is straightforward, finding ones that should be decommissioned is trickier. The usual approach to hold resource owners accountable for budgets isn't working since they're not exceeding them, but they are wasting money on these unnecessary resources. I'm interested in hearing other strategies to tackle this issue!
5 Answers
One effective method is to use the Azure Cost Optimization Workbook. It can help you pinpoint orphaned resources quickly. Create a list of these resources and develop a plan for their deletion. It gives you the visibility needed to start cleaning house!
I still think you should make the subscription or resource group owners accountable, but without the budget pressure. Start regular reviews and send updates to resource owners so they can confirm the resource's necessity. If someone claims it's needed, keep it; if they want to ditch it, do it! This makes everyone more aware of what they have.
Don't forget to leverage Azure’s tools and community workbooks to track orphaned resources and streamline optimizations. We've found that tagging helps a lot to keep track of "what is this for?" When it comes to legacy resources, although the costs are often negligible, they can pile up, so it’s better to stay on top of them!
This really is an operational issue; if the teams can’t tell if a resource is in use, relying solely on FinOps tools won't fix the bigger problem. It also might be beneficial to have periodic reviews where resource owners need to justify the need for each resource. If they can’t, that’s a solid case for removal!
Finding an automated tool that solves all these problems is tough. Most tools will flag potential candidates, and it really requires someone familiar with the environment to assess them. Regular check-ins are essential to keep operations smooth. I've been developing a tool that flags underutilized VMs and tracks changes over time. It’s still in beta, but it might help streamline some processes if you're interested!
Totally agree, accountability is key! If they can't keep track of what they own, it’s going to be a perpetual issue.