Hey everyone! I'm new to my role as a cloud analyst and tasked with finding cost optimization opportunities in our AWS setup. We've got two accounts—one for production and another for CI/CD and internal systems. We're using AWS Control Tower for governance, and our key services include EC2, RDS, S3, Lambda, Elastic Beanstalk, ECS, CloudFront, and EventBridge. Security Hub and AWS Config are enabled, and we use IAM with least privilege.
So far, I've mapped our environment using AWS CLI and identified some potential cost-saving opportunities, such as over-provisioned EC2 instances in non-production environments; idle RDS instances; and unused Elastic Beanstalk environments and S3 buckets. I've proposed a phased plan starting with scheduling EC2 shutdowns during non-business hours, right-sizing instances, and removing idle resources. I want to get feedback on this approach before moving forward.
Here are my specific questions:
1. Does it make sense to take this phased approach as a newbie in a critical environment?
2. Are there common mistakes to avoid when right-sizing or deleting old resources?
3. How do you keep the team aligned before removing resources?
4. Can I safely enable automation tools like Instance Scheduler in a Control Tower setup?
5. Any recommended practices for tracking savings in FinOps?
I'm focusing on low-impact changes first to build trust with the team. Any advice, tips, or war stories would be fantastic! Thanks!
1 Answer
Have you taken a close look at your billing? Understanding the major cost areas is crucial. Using savings plans effectively and checking for any VPC endpoints can provide additional savings. Additionally, focus on identifying unused resources like EBS volumes and consider implementing storage tiering. It’s wise to start with the low-risk opportunities that won’t impact your environment before moving on to those that might save you 10-20%.
Thanks for the feedback! This is just the kind of guidance I need as I’m getting my feet wet. I’m already working on mapping the environment and have access to Cost Explorer. I’ve identified some over-provisioned EC2 instances and idle resources, and I’m planning a phased approach to tackle them efficiently.